Run 3Dtex-weigthed NH with internal cross-validation of parameters
# 1) From 100% dataset, Create train/validation (80%) / heldout (20%) partitions
sqlite <- dbDriver("SQLite")
conn <- dbConnect(sqlite, "textureUpdatedFeatures.db")
# 2) all T1w features
lesionsQuery <- dbGetQuery(conn, "SELECT *
FROM stage1features
INNER JOIN lesion ON (stage1features.lesion_id = lesion.lesion_id)
INNER JOIN f_T2 ON (stage1features.lesion_id = f_T2.lesion_id)")
# For bootstrap resampling, createResample is used
# Randomization is done at the patient level, so that bootstrapping preserves lesion independence
lesionsQuery = subset(lesionsQuery, lesion_label != "fociB" & lesion_label != "fociM" ) # exclude foci at this point
id_cad_pts = lesionsQuery$cad_pt_no_txt
uniq_cad = unique(lesionsQuery$cad_pt_no_txt)
npatients = 1:length(uniq_cad)
# when y is a factor in an attempt to balance the class distributions within the splits.
# The names of the list objects will denote the fold membership using the pattern
# resamples." meaning the ith section (of k) of the jth cross-validation set (of times).
set.seed(1234)
npatients = length(uniq_cad)
kfcvpartitionsetD <- createFolds(y = 1:length(uniq_cad),## the outcome data are needed
k = 10,
list = TRUE)
## using 3Dtexture first + Boosting
perf_imgT2 = data.frame();
perf_allT2 = data.frame();
perf_imgT1 = data.frame();
perf_all = data.frame();
## holders for reature rankings
imgT2featsel = data.frame()
allT2featsel = data.frame()
imgT1featsel = data.frame()
allfeatsel = data.frame()
cvauc_imgT2 = c()
cvauc_allT2 = c()
cvauc_imgT1 = c()
cvauc_all = c()
# perform k-fold-out
for(k in 1:10){ # 1:10f cv
## Create folds leave-one-patient-out
allfT2 = read3Dtex_T2uniqcad_parti(id_cad_pts, uniq_cad, kfcvpartitionsetD, 10, k)
allfT1 = read3Dtex_T1uniqcad_parti(id_cad_pts, uniq_cad, kfcvpartitionsetD, 10, k)
allfT1T2 = read3Dtex_T1T2uniqcad_parti(id_cad_pts, uniq_cad, kfcvpartitionsetD, 10, k)
## formant
T2train = allfT2[[1]]; T2traininfo = allfT2[[5]]; T2trainids = T2traininfo$lesion_id;
T2test = allfT2[[2]]; T2testinfo = allfT2[[6]]; T2testids = T2testinfo$lesion_id;
T1train = allfT1[[1]]; T1traininfo = allfT1[[5]]; T1trainids = T1traininfo$lesion_id;
T1test = allfT1[[2]]; T1testinfo = allfT1[[6]]; T1testids = T1testinfo$lesion_id;
T1T2train = allfT1T2[[1]]; T1T2traininfo = allfT1T2[[5]]; T1T2trainids = T1T2traininfo$lesion_id;
T1T2test = allfT1T2[[2]]; T1T2testinfo = allfT1T2[[6]]; T1T2testids = T1T2testinfo$lesion_id;
# remove radiologist based BIRADS category and measured muscle-to-lesion SI
# add predicted T2w features
T2LMSIR = getid_predLMSIR(LMSIR_lop, T2trainids)
T2wSI = getid_predT2wSI(perfT2wSI_lop, T2trainids)
imgT2train = T2train[,-c(2,5,ncol(T2train))] # exlude orig_label
wpredT2train = cbind(imgT2train, LMSIR_predicted=T2LMSIR$LMSIR_predicted, T2wSI_predicted=T2wSI$T2wSI_predicted)
wpredT2train$T2wSI_predicted = as.factor(wpredT2train$T2wSI_predicted)
T1train = T1train[,-c(ncol(T1train))]
# remove radiologist based BIRADS category and measured muscle-to-lesion SI
# add predicted T2w features
T1T2LMSIR = getid_predLMSIR(LMSIR_lop, T1T2trainids)
T1T2wSI = getid_predT2wSI(perfT2wSI_lop, T1T2trainids)
########## consider differneces
imgT1T2train = T1T2train[,-c(199,202,ncol(T1T2train))]
wpredT1T2train = cbind(imgT1T2train, LMSIR_predicted=T1T2LMSIR$LMSIR_predicted, T2wSI_predicted=T1T2wSI$T2wSI_predicted)
wpredT1T2train$T2wSI_predicted = as.factor(wpredT1T2train$T2wSI_predicted)
# with datasets: T2train, wpredT2train, T1train, T1T2train, wpredT1T2train
selrrfimgT2 = RRF_featsel(imgT2train, "imgT2")
selrrfallT2 = RRF_featsel(wpredT2train, "allT2")
selrrfimgT1 = RRF_featsel(T1train, "imgT1")
selrrfall = RRF_featsel(wpredT1T2train, "all")
## group with all of the features spaces combined, most contributing T2w feature
imgT2featsel = rbind(imgT2featsel, cbind(selrrfimgT2, kfcv=k) )
allT2featsel = rbind(allT2featsel, cbind(selrrfallT2, kfcv=k) )
imgT1featsel = rbind(imgT1featsel, cbind(selrrfimgT1, kfcv=k) )
allfeatsel = rbind(allfeatsel, cbind(selrrfall, kfcv=k) )
##################
# Define datasets
##################
# define datasets: imgT2wfeatures allT2wfeatures, imgT1wfeatures, allfeatures
imgT2features = imgT2train[,c("lesion_label", selrrfimgT2$selfeat)]
allT2features = wpredT2train[,c("lesion_label",selrrfallT2$selfeat)]
imgT1features = T1train[,c("lesion_label",selrrfimgT1$selfeat)]
allfeatures = wpredT1T2train[, c("lesion_label",selrrfall$selfeat)]
##################
# Get Test info data
##################
dfinfo = cbind(T2testinfo[,c(1,3,6,24:26)],
find_t2_signal_int=T2test$find_t2_signal_int)
print(dfinfo)
## apend LMSIR and T2wSI in case is used by classifier
testLMSIR = getid_predLMSIR(LMSIR_lop, T2testids)
testT2wSI = getid_predT2wSI(perfT2wSI_lop, T2testids)
T2test = cbind(T2test, LMSIR_predicted=testLMSIR$LMSIR_predicted,
T2wSI_predicted=testT2wSI$T2wSI_predicted)
## for T1T2
## apend LMSIR and T2wSI in case is used by classifier
testLMSIR = getid_predLMSIR(LMSIR_lop, T1T2testids)
testT2wSI = getid_predT2wSI(perfT2wSI_lop, T1T2testids)
T1T2test = cbind(T1T2test, LMSIR_predicted=testLMSIR$LMSIR_predicted, T2wSI_predicted=testT2wSI$T2wSI_predicted)
##################
# Build final classifiers
##################
# data = imgT2features,
cat("\n============ bagging trees treedata_imgT2 \n")
# train trees
treedata_imgT2 <- NH_looforestTrain_wcv(imgT2features, T2test, c(1,2,3))
######## data = allT2features,
cat("\n============ bagging trees treedata_allT2 \n")
# train trees
if(k==5){allT2features=allT2features[,-c(2)]}
treedata_allT2 <- NH_looforestTrain_wcv(allT2features, T2test, c(1,2,3))
####### data = imgT1features,
cat("\n============ bagging trees treedata_imgT1 \n")
# train trees
treedata_imgT1 <- NH_looforestTrain_wcv(imgT1features, T1test, c(1,2,3,5))
####### data = allfeatures,
cat("\n============ bagging trees treedata_all \n")
# train trees
treedata_all <- NH_looforestTrain_wcv(allfeatures, T1T2test, c(1,2,3,5))
##################
### predict for each classifier
##################
## for treedata_imgT2
rules = data.frame(C=treedata_imgT2$testperf$testpred_NH, NC=1-treedata_imgT2$testperf$testpred_NH)
rules$pred = apply(rules, 1, which.max)
perfcv_imgT2 = data.frame(id=treedata_imgT2$testperf$ids,
C=treedata_imgT2$testperf$testpred_NH,
NC=1-treedata_imgT2$testperf$testpred_NH,
pred=ifelse(rules$pred==1,"C","NC"), obs=treedata_imgT2$testperf$labelstest)
auc_imgT2 = roc(perfcv_imgT2$obs, perfcv_imgT2$C)
perf_imgT2 = rbind(perf_imgT2, perfcv_imgT2 )
print(head(perfcv_imgT2))
cvauc_imgT2 = c(cvauc_imgT2, auc_imgT2$auc)
# for treedata_allT2
rules = data.frame(C=treedata_allT2$testperf$testpred_NH, NC=1-treedata_allT2$testperf$testpred_NH)
rules$pred = apply(rules, 1, which.max)
perfcv_allT2 = data.frame(id=treedata_allT2$testperf$ids,
C=treedata_allT2$testperf$testpred_NH,
NC=1-treedata_allT2$testperf$testpred_NH,
pred=ifelse(rules$pred==1,"C","NC"), obs=treedata_allT2$testperf$labelstest)
auc_allT2 = roc(perfcv_allT2$obs, perfcv_allT2$C)
perf_allT2 = rbind(perf_allT2, perfcv_allT2)
print(head(perfcv_allT2))
cvauc_allT2 = c(cvauc_allT2, auc_allT2$auc)
## for treedata_imgT1
rules = data.frame(C=treedata_imgT1$testperf$testpred_NH, NC=1-treedata_imgT1$testperf$testpred_NH)
rules$pred = apply(rules, 1, which.max)
perfcv_imgT1 = data.frame(id=treedata_imgT1$testperf$ids,
C=treedata_imgT1$testperf$testpred_NH,
NC=1-treedata_imgT1$testperf$testpred_NH,
pred=ifelse(rules$pred==1,"C","NC"), obs=treedata_imgT1$testperf$labelstest)
auc_imgT1 = roc(perfcv_imgT1$obs, perfcv_imgT1$C)
perf_imgT1 = rbind(perf_imgT1, perfcv_imgT1)
print(head(perfcv_imgT1))
cvauc_imgT1 = c(cvauc_imgT1, auc_imgT1$auc)
# for treedata_all
rules = data.frame(C=treedata_all$testperf$testpred_NH, NC=1-treedata_all$testperf$testpred_NH)
rules$pred = apply(rules, 1, which.max)
perfcv_all = data.frame(id=treedata_all$testperf$ids,
C=treedata_all$testperf$testpred_NH,
NC=1-treedata_all$testperf$testpred_NH,
pred=ifelse(rules$pred==1,"C","NC"), obs=treedata_all$testperf$labelstest)
auc_all = roc(perfcv_all$obs, perfcv_all$C)
perf_all = rbind(perf_all, perfcv_all)
print(head(perfcv_all))
cvauc_all = c(cvauc_all, auc_all$auc)
# AUC
rocperf_imgT2 = roc(perf_imgT2$obs, perf_imgT2$C)
print(rocperf_imgT2)
rocperf_allT2 = roc(perf_allT2$obs, perf_allT2$C)
print(rocperf_allT2)
rocperf_imgT1 = roc(perf_imgT1$obs, perf_imgT1$C)
print(rocperf_imgT1)
rocperf_all = roc(perf_all$obs, perf_all$C)
print(rocperf_all)
# plot every 10 patients
## plot ROCs each pass individually in l-o-p heldout test cases
par(mfrow=c(1,1))
n=15
colors = rainbow(n, s = 1, v = 1, start = 0, end = max(1, n - 1)/n, alpha = 1)
# plot 1/4
p1 = calcAUC_plot(perf_imgT2$obs, perf_imgT2$C,
xptext=0.45, yptext=0.75 ,colors[2], atitle="")
par(new=TRUE)
p2 = calcAUC_plot(perf_allT2$obs, perf_allT2$C,
xptext=0.55, yptext=0.65 ,colors[9], atitle="")
par(new=TRUE)
p3 = calcAUC_plot(perf_imgT1$obs, perf_imgT1$C,
xptext=0.65, yptext=0.55 ,colors[11], atitle="")
par(new=TRUE)
p4 = calcAUC_plot(perf_all$obs, perf_all$C,
xptext=0.75, yptext=0.45 ,colors[14],
atitle=paste0("ROCs 10f-patient out cv test k-fold= ",k))
legend("bottomright",
legend = c(paste0("imgT2w"),
paste0("imgT2w+predT2w"),
paste0("imgT1w"),
paste0("imgT1+imgT2w+predT2w")),
col = c(colors[2],colors[9],colors[11],colors[14]), lwd = 2)
# save current state k patient out
save.image(paste0("Outputs/weigNH_addeddiagvalue_3Dtexbagg_cv",k,".RData"))
}
## massB massM nonmassB nonmassM
## 216 153 131 67
## massB massM nonmassB nonmassM
## 24 15 11 10
## massB massM nonmassB nonmassM
## 216 153 131 67
## massB massM nonmassB nonmassM
## 24 15 11 10
## massB massM nonmassB nonmassM
## 216 153 131 67
## massB massM nonmassB nonmassM
## 24 15 11 10
## RRF 1.6
## Type rrfNews() to see new features/changes/bug fixes.
##
## Attaching package: 'RRF'
##
## The following object is masked from 'package:ranger':
##
## importance
##
## The following object is masked from 'package:ggplot2':
##
## margin
## -0.0201005 0.05
## Selected features for group: MeanDecreaseGini imgT2
## =========NULL
## [1] "T2RGH_var" "ave_T210"
## [3] "T2texture_inversediffmoment_nondir" "T2RGH_mean"
## [5] "ave_T27" "T2texture_entropy_nondir"
## [7] "T2max_F_r_i" "ave_T22"
## [9] "T2texture_correlation_nondir" "T2kurt_F_r_i"
## [11] "ave_T211" "ave_T25"
## [13] "ave_T212" "ave_T28"
## [15] "ave_T23" "T2min_F_r_i"
## [17] "ave_T24" "T2texture_sumvariance_nondir"
## [19] "T2grad_margin_var" "T2texture_variance_nondir"
## [21] "ave_T214" "ave_T29"
## [23] "T2grad_margin" "T2texture_diffvariance_nondir"
## [25] "ave_T215" "ave_T219"
## [27] "T2skew_F_r_i" "T2mean_F_r_i"
## [29] "T2_lesionSI"
## 0.04807692 0.05
## Selected features for group: MeanDecreaseGini allT2
## =========NULL
## [1] "T2RGH_var" "T2texture_entropy_nondir"
## [3] "T2texture_correlation_nondir" "T2RGH_mean"
## [5] "ave_T210" "T2texture_sumaverage_nondir"
## [7] "T2_lesionSIstd" "T2wSI_predicted"
## [9] "T2skew_F_r_i" "T2texture_diffvariance_nondir"
## [11] "ave_T27" "T2grad_margin_var"
## [13] "LMSIR_predicted" "ave_T212"
## [15] "ave_T25" "ave_T20"
## [17] "ave_T217" "ave_T28"
## [19] "ave_T24" "ave_T218"
## [21] "ave_T219" "ave_T216"
## [23] "ave_T29" "T2kurt_F_r_i"
## [25] "ave_T211" "T2texture_energy_nondir"
## [27] "ave_T215" "ave_T22"
## [29] "ave_T23" "T2grad_margin"
## [31] "ave_T26" "T2texture_inversediffmoment_nondir"
## [33] "T2_lesionSI" "ave_T21"
## [35] "T2texture_sumentropy_nondir" "ave_T213"
## [37] "T2texture_sumvariance_nondir" "T2min_F_r_i"
## 0.07894737 0.05
## -0.08571429 0.05
## Selected features for group: MeanDecreaseGini imgT1
## =========NULL
## [1] "irregularity" "SER_countor"
## [3] "texture_variance_nondir_post2" "Slope_ini_inside"
## [5] "mean_F_r_i" "V15"
## [7] "alpha_inside" "min_F_r_i"
## [9] "Vr_post_1_inside" "texture_sumaverage_nondir_post2"
## [11] "texture_sumentropy_nondir_post3" "V0"
## [13] "texture_inversediffmoment_nondir_post1" "iiiMax_Margin_Gradient"
## [15] "V5" "dce3SE8"
## [17] "lateSE0" "dce2SE18"
## [19] "texture_correlation_nondir_post2" "dce3SE9"
## [21] "dce3SE2" "Vr_decreasingRate_countor"
## [23] "dce3SE3" "V19"
## [25] "Vr_decreasingRate_inside" "texture_sumaverage_nondir_post3"
## [27] "dce2SE5" "dce3SE7"
## [29] "texture_entropy_nondir_post3" "earlySE11"
## [31] "earlySE12" "max_RGH_mean_k"
## [33] "texture_sumentropy_nondir_post4" "A_inside"
## 0.01986755 0.05
## Selected features for group: MeanDecreaseGini all
## =========NULL
## [1] "texture_variance_nondir_post1" "circularity"
## [3] "texture_energy_nondir_post4" "earlySE12"
## [5] "V12" "texture_contrast_nondir_post3"
## [7] "V13" "max_F_r_i"
## [9] "V11" "iiMin_change_Variance_uptake"
## [11] "Vr_increasingRate_inside" "V5"
## [13] "lateSE17" "lateSE6"
## [15] "ave_T213" "V17"
## [17] "V14" "texture_sumaverage_nondir_post4"
## [19] "dce3SE18" "UptakeRate_inside"
## [21] "T2mean_F_r_i" "T2texture_contrast_nondir"
## [23] "lateSE0" "maxCr_countor"
## [25] "ave_T219" "A_countor"
## [27] "T2grad_margin_var" "ave_T216"
## [29] "Vr_post_1_countor" "beta_inside"
## [31] "dce3SE4" "V18"
## [33] "dce3SE16" "peakCr_inside"
## [35] "V2" "ave_T215"
## [37] "A_inside"
## lesion_id cad_pt_no_txt exam_a_number_txt BIRADS lesion_label
## 1 1 0002 6745896 4 nonmassM
## 32 32 0173 5123923 4 nonmassB
## 40 40 0190 6760690 4 massM
## 41 41 0190 6760690 4 nonmassM
## 51 51 0205 5085133 4 massB
## 103 103 0553 6687000 2 massB
## 104 104 0561 4668611 4 massB
## 116 116 0606 6781309 4 massB
## 117 117 0608 5094101 4 massB
## 134 134 0672 4899757 5 nonmassM
## 152 152 0696 6983274 4 massB
## 153 153 0700 4660805 5 massM
## 165 165 0718 4962581 4 massB
## 179 179 0730 5009497 5 massM
## 185 185 0740 4842984 4 nonmassB
## 216 216 0779 4934249 5 massB
## 217 217 0779 4934249 5 massM
## 227 227 0796 860773 4 nonmassB
## 228 228 0799 5372294 4 massB
## 229 229 0802 4600874 4 massM
## 232 232 0807 5235491 5 nonmassM
## 284 284 0871 5130094 4 massM
## 285 285 0871 5130094 4 massM
## 286 286 0871 5130094 4 massM
## 301 301 0885 6747175 4 nonmassB
## 309 309 0918 6976567 4 massB
## 332 332 0993 6979299 4 massM
## 367 367 1086 7173349 6 nonmassB
## 378 378 2016 7052211 4 massB
## 380 380 2023 5141524 6 massM
## 399 399 2059 7749617 4 nonmassB
## 413 413 3005 4974097 3 nonmassB
## 414 414 3005 6757337 3 nonmassM
## 415 415 3005 5057668 2 nonmassB
## 416 416 3005 6757337 4 nonmassM
## 422 422 3020 7395195 4 massB
## 424 424 3023 7106703 6 massM
## 451 451 3065 7037223 4 massB
## 452 452 3065 7037223 4 massB
## 453 453 3070 7085188 4 massB
## 465 465 3081 7041435 5 nonmassB
## 466 466 3081 7041435 5 nonmassM
## 476 476 4003 7056445 4 nonmassB
## 477 477 4003 7056445 4 nonmassB
## 492 492 4029 7633460 4 massB
## 493 493 4029 7633460 4 massB
## 503 503 4047 7009608 4 massB
## 514 514 6015 5082265 6 massM
## 515 515 6015 5082265 6 nonmassM
## 518 518 6018 5088825 5 nonmassM
## 521 521 6021 4798692 4 massB
## 534 534 6029 5083338 6 massB
## 535 535 6029 5083338 6 massB
## 536 536 6029 6772981 4 massB
## 538 538 6034 4997881 6 massM
## 539 539 6034 4997881 6 massM
## 540 540 6034 4997881 6 nonmassM
## 593 593 7008 6875110 6 massM
## 602 602 7045 6760802 4 massB
## 631 631 7190 7013378 3 massB
## lesion_diagnosis find_t2_signal_int
## 1 InsituDuctal None
## 32 DUCT PAPILLOMA None
## 40 InvasiveDuctal None
## 41 InvasiveDuctal None
## 51 FIBROEPITHELIAL Hyperintense
## 103 BENIGN BREAST TISSUE Hyperintense
## 104 FIBROADENOMA Hypointense or not seen
## 116 ATYPICAL DUCTAL HYPERPLASIA Hyperintense
## 117 BENIGN BREAST TISSUE Hyperintense
## 134 InsituDuctal None
## 152 BENIGN BREAST TISSUE Hypointense or not seen
## 153 InvasiveDuctal None
## 165 FIBROCYSTIC None
## 179 InvasiveDuctal None
## 185 SCLEROSING INTRADUCTAL PAPILLOMA Slightly hyperintense
## 216 SCLEROSING ADENOSIS None
## 217 InvasiveDuctal None
## 227 ATYPICAL LOBULAR HYPERPLASIA None
## 228 FIBROCYSTIC Slightly hyperintense
## 229 InvasiveDuctal None
## 232 InsituDuctal None
## 284 InvasiveLobular None
## 285 InsituDuctal None
## 286 InvasiveLobular None
## 301 FIBROCYSTIC Hypointense or not seen
## 309 FIBROADENOMA Hyperintense
## 332 PHYLLODES TUMOR Hyperintense
## 367 ATYPICAL DUCTAL HYPERPLASIA None
## 378 FIBROADENOMA Hyperintense
## 380 InvasiveDuctal None
## 399 COLUMNAR CELL CHANGES None
## 413 BENIGN BREAST TISSUE None
## 414 InsituDuctal Hypointense or not seen
## 415 FIBROCYSTIC None
## 416 InsituDuctal Hypointense or not seen
## 422 STROMAL HYPERPLASIA Hypointense or not seen
## 424 InvasiveDuctal None
## 451 ADENOSIS Hypointense or not seen
## 452 ADENOSIS Hypointense or not seen
## 453 DUCTAL HYPERPLASIA WITHOUT ATYPIA Hypointense or not seen
## 465 COLUMNAR CELL CHANGES Hypointense or not seen
## 466 InsituDuctal Hypointense or not seen
## 476 FLORID DUCTAL HYPERPLASIA None
## 477 FLORID DUCTAL HYPERPLASIA None
## 492 InsituLobular None
## 493 InsituLobular None
## 503 FIBROCYSTIC Hypointense or not seen
## 514 InvasiveDuctal None
## 515 InvasiveDuctal None
## 518 InvasiveLobular None
## 521 ATYPICAL LOBULAR HYPERPLASIA None
## 534 FIBROADENOMA Hypointense or not seen
## 535 FIBROADENOMA Hypointense or not seen
## 536 FIBROADENOMA Hypointense or not seen
## 538 InvasiveDuctal Hyperintense
## 539 InvasiveDuctal Hyperintense
## 540 InvasiveDuctal None
## 593 InvasiveLobular Hyperintense
## 602 BENIGN BREAST TISSUE Hypointense or not seen
## 631 COLUMNAR CELL CHANGES Hyperintense
##
## ============ bagging trees treedata_imgT2
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6714697 0.6211429 0.6804046
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.7686141 0.6314286 0.7346014
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8179067 0.6268571 0.7527174
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 4 1 20 0.6714697 0.6268571 0.6837258
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 2 20 0.7592088 0.6405714 0.7471316
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 3 20 0.8124574 0.636 0.7727959
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 7 1 15 0.6714697 0.6268571 0.6837258
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 8 2 15 0.7625164 0.6617143 0.7673611
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 3 15 0.8108528 0.7148571 0.7999698
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 10 1 10 0.6714697 0.6268571 0.6837258
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 11 2 10 0.7479041 0.6537143 0.7424517
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 12 3 10 0.7917933 0.6274286 0.7400362
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 5 0.6714697 0.6211429 0.6804046
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 5 0.7595887 0.6205714 0.7243357
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 5 0.7920422 0.6542857 0.750151
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6714697 0.6211429 0.6804046
## 2 2 25 0.7686141 0.6314286 0.7346014
## 3 3 25 0.8179067 0.6268571 0.7527174
## 4 1 20 0.6714697 0.6268571 0.6837258
## 5 2 20 0.7592088 0.6405714 0.7471316
## 6 3 20 0.8124574 0.6360000 0.7727959
## 7 1 15 0.6714697 0.6268571 0.6837258
## 8 2 15 0.7625164 0.6617143 0.7673611
## 9 3 15 0.8108528 0.7148571 0.7999698
## 10 1 10 0.6714697 0.6268571 0.6837258
## 11 2 10 0.7479041 0.6537143 0.7424517
## 12 3 10 0.7917933 0.6274286 0.7400362
## 13 1 5 0.6714697 0.6211429 0.6804046
## 14 2 5 0.7595887 0.6205714 0.7243357
## 15 3 5 0.7920422 0.6542857 0.7501510
## maxinter nodesize rocTrain rocValid rocTest
## 9 3 15 0.8108528 0.7148571 0.7999698
##
## Observation 1 has a predicted value 0.458
## since this is the weighted average response across the 9 nodes it is a member of:
##
## 1) Node 39, containing 242 training observations, with node mean 0.302 and weight 0.218 :
## T2texture_entropy_nondir <= 3.5
## T2texture_variance_nondir <= 360
## T2texture_correlation_nondir <= 0.26
##
## 2) Node 15, containing 35 training observations, with node mean 0.429 and weight 0.19 :
## T2texture_correlation_nondir <= 0.26
## T2texture_entropy_nondir <= 3.5
## T2_lesionSI <= 60
##
## 3) Node 28, containing 100 training observations, with node mean 0.66 and weight 0.178 :
## 350 <= T2RGH_var
## 50 <= ave_T211 <= 140
##
## 4) Node 41, containing 256 training observations, with node mean 0.516 and weight 0.138 :
## T2min_F_r_i <= 8.5
## 56 <= ave_T210
## T2grad_margin_var <= 3800
##
## 5) Node 42, containing 270 training observations, with node mean 0.359 and weight 0.0991 :
## T2texture_correlation_nondir <= 0.26
## 67 <= ave_T29
## T2min_F_r_i <= 59
##
## 6) Node 27, containing 88 training observations, with node mean 0.455 and weight 0.0683 :
## 350 <= T2RGH_var
## T2texture_entropy_nondir <= 3.2
## T2RGH_mean <= 41
##
## 7) Node 31, containing 112 training observations, with node mean 0.621 and weight 0.0599 :
## 350 <= T2RGH_var
## 46 <= ave_T211 <= 150
##
## 8) Node 43, containing 386 training observations, with node mean 0.371 and weight 0.0361 :
## T2texture_entropy_nondir <= 3.5
## T2texture_diffvariance_nondir <= 390
## 20 <= ave_T211
##
## 9) Node 44, containing 563 training observations, with node mean 0.388 and weight 0.01 :
## ROOT NODE
##
## ============ bagging trees treedata_allT2
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6892913 0.7205714 0.7651976
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.758508 0.6222857 0.7270517
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8155358 0.7005714 0.8012158
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 4 1 20 0.6892913 0.7142857 0.7621581
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 2 20 0.7597131 0.6325714 0.7378419
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 3 20 0.7977535 0.6 0.7430091
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 7 1 15 0.6892913 0.7205714 0.7651976
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 8 2 15 0.7590844 0.6857143 0.7662614
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 3 15 0.8228386 0.5617143 0.7291793
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 10 1 10 0.6892913 0.7205714 0.7651976
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 11 2 10 0.7634595 0.7074286 0.7919453
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 12 3 10 0.808842 0.7022857 0.7951368
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 5 0.6892913 0.7205714 0.7651976
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 5 0.7608069 0.6617143 0.7545593
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 5 0.8106694 0.6377143 0.7569909
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6892913 0.7205714 0.7651976
## 2 2 25 0.7585080 0.6222857 0.7270517
## 3 3 25 0.8155358 0.7005714 0.8012158
## 4 1 20 0.6892913 0.7142857 0.7621581
## 5 2 20 0.7597131 0.6325714 0.7378419
## 6 3 20 0.7977535 0.6000000 0.7430091
## 7 1 15 0.6892913 0.7205714 0.7651976
## 8 2 15 0.7590844 0.6857143 0.7662614
## 9 3 15 0.8228386 0.5617143 0.7291793
## 10 1 10 0.6892913 0.7205714 0.7651976
## 11 2 10 0.7634595 0.7074286 0.7919453
## 12 3 10 0.8088420 0.7022857 0.7951368
## 13 1 5 0.6892913 0.7205714 0.7651976
## 14 2 5 0.7608069 0.6617143 0.7545593
## 15 3 5 0.8106694 0.6377143 0.7569909
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8155358 0.7005714 0.8012158
##
## Observation 1 has a predicted value 0.518
## since this is the weighted average response across the 10 nodes it is a member of:
##
## 1) Node 36, containing 67 training observations, with node mean 0.716 and weight 0.27 :
## 350 <= T2RGH_var
## T2RGH_mean <= 41
## ave_T211 <= 120
##
## 2) Node 58, containing 0.5 training observations, with node mean 0.593 and weight 0.189 :
## T2wSI_predicted = None
## 60 <= ave_T210
## T2min_F_r_i <= 3.5
##
## 3) Node 43, containing 0.5 training observations, with node mean 0.302 and weight 0.173 :
## T2wSI_predicted = None
## T2texture_entropy_nondir <= 3.2
## 2 <= LMSIR_predicted
##
## 4) Node 61, containing 0.5 training observations, with node mean 0.397 and weight 0.113 :
## 22 <= T2RGH_mean
## T2texture_correlation_nondir <= 0.26
## T2wSI_predicted in {Hypointense or not seen,None,Slightly hyperintense}
##
## 5) Node 8, containing 15 training observations, with node mean 0.467 and weight 0.063 :
## T2min_F_r_i <= 8.5
## T2texture_correlation_nondir <= 0.15
## 0.19 <= T2texture_inversediffmoment_nondir
##
## 6) Node 57, containing 0.5 training observations, with node mean 0.45 and weight 0.0585 :
## T2texture_entropy_nondir <= 3.5
## T2texture_sumvariance_nondir <= 1200
## T2wSI_predicted = None
##
## 7) Node 42, containing 90 training observations, with node mean 0.311 and weight 0.0548 :
## T2texture_entropy_nondir <= 3.5
## 0.13 <= T2texture_inversediffmoment_nondir
## T2texture_diffvariance_nondir <= 58
##
## 8) Node 49, containing 117 training observations, with node mean 0.627 and weight 0.0519 :
## 350 <= T2RGH_var
## 54 <= ave_T210
## ave_T211 <= 140
##
## 9) Node 63, containing 378 training observations, with node mean 0.374 and weight 0.0177 :
## T2texture_entropy_nondir <= 3.5
## T2texture_sumvariance_nondir <= 1100
## ave_T219 <= 340
##
## 10) Node 64, containing 563 training observations, with node mean 0.388 and weight 0.01 :
## ROOT NODE
##
## ============ bagging trees treedata_imgT1
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.8137346 0.6365714 0.7631965
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.8376015 0.632 0.7859238
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8647302 0.608 0.7699413
## maxinter: 5 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 4 5 25 0.9183849 0.696 0.8398827
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 1 20 0.8137477 0.6365714 0.7631965
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 2 20 0.8372216 0.6022857 0.7620235
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 7 3 20 0.8672976 0.6457143 0.7970674
## maxinter: 5 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 8 5 20 0.9031045 0.656 0.8042522
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 1 15 0.8137346 0.6365714 0.7631965
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 10 2 15 0.8404899 0.616 0.768915
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 11 3 15 0.8752489 0.6662857 0.8111437
## maxinter: 5 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 12 5 15 0.9000131 0.6457143 0.8111437
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 10 0.8137346 0.6365714 0.7631965
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 10 0.8382237 0.624 0.7766862
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 10 0.8677495 0.6034286 0.7983871
## maxinter: 5 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 16 5 10 0.9083639 0.6525714 0.8246334
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 17 1 5 0.8137477 0.6365714 0.7631965
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 18 2 5 0.8435093 0.6194286 0.7636364
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 19 3 5 0.8658501 0.6468571 0.7881232
## maxinter: 5 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 20 5 5 0.9012706 0.6662857 0.8230205
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.8137346 0.6365714 0.7631965
## 2 2 25 0.8376015 0.6320000 0.7859238
## 3 3 25 0.8647302 0.6080000 0.7699413
## 4 5 25 0.9183849 0.6960000 0.8398827
## 5 1 20 0.8137477 0.6365714 0.7631965
## 6 2 20 0.8372216 0.6022857 0.7620235
## 7 3 20 0.8672976 0.6457143 0.7970674
## 8 5 20 0.9031045 0.6560000 0.8042522
## 9 1 15 0.8137346 0.6365714 0.7631965
## 10 2 15 0.8404899 0.6160000 0.7689150
## 11 3 15 0.8752489 0.6662857 0.8111437
## 12 5 15 0.9000131 0.6457143 0.8111437
## 13 1 10 0.8137346 0.6365714 0.7631965
## 14 2 10 0.8382237 0.6240000 0.7766862
## 15 3 10 0.8677495 0.6034286 0.7983871
## 16 5 10 0.9083639 0.6525714 0.8246334
## 17 1 5 0.8137477 0.6365714 0.7631965
## 18 2 5 0.8435093 0.6194286 0.7636364
## 19 3 5 0.8658501 0.6468571 0.7881232
## 20 5 5 0.9012706 0.6662857 0.8230205
## maxinter nodesize rocTrain rocValid rocTest
## 4 5 25 0.9183849 0.696 0.8398827
##
## Observation 1 has a predicted value 0.427
## since this is the weighted average response across the 10 nodes it is a member of:
##
## 1) Node 81, containing 58 training observations, with node mean 0.241 and weight 0.269 :
## 0.71 <= SER_countor
## mean_F_r_i <= 790
## irregularity <= 0.96
## 1.8 <= texture_sumentropy_nondir_post4
##
## 2) Node 87, containing 65 training observations, with node mean 0.231 and weight 0.176 :
## texture_variance_nondir_post2 <= 310
## 0.71 <= SER_countor
## irregularity <= 0.92
## 3.5 <= V5
##
## 3) Node 86, containing 64 training observations, with node mean 0.859 and weight 0.12 :
## 1.6 <= dce2SE5
## 200 <= texture_variance_nondir_post2
##
## 4) Node 80, containing 55 training observations, with node mean 0.655 and weight 0.111 :
## irregularity <= 0.98
## 0.84 <= SER_countor
## 6.5 <= V19
##
## 5) Node 89, containing 70 training observations, with node mean 0.3 and weight 0.105 :
## 180 <= texture_variance_nondir_post2
## 0.51 <= Slope_ini_inside
## irregularity <= 0.97
## mean_F_r_i <= 1100
## 1.8 <= texture_sumentropy_nondir_post3
##
## 6) Node 99, containing 128 training observations, with node mean 0.453 and weight 0.0717 :
## texture_variance_nondir_post2 <= 250
## 0.6 <= Slope_ini_inside
## 0.88 <= irregularity
## 0.13 <= Vr_post_1_inside
## 0.095 <= iiiMax_Margin_Gradient
##
## 7) Node 28, containing 16 training observations, with node mean 0.438 and weight 0.0701 :
## texture_variance_nondir_post2 <= 250
## irregularity <= 0.93
## 1.7 <= earlySE11
##
## 8) Node 44, containing 21 training observations, with node mean 0.714 and weight 0.0647 :
## irregularity <= 0.98
## 0.56 <= alpha_inside
## 0.13 <= Vr_post_1_inside
## 0.64 <= texture_correlation_nondir_post2
##
## 9) Node 110, containing 567 training observations, with node mean 0.388 and weight 0.01 :
## ROOT NODE
##
## 10) Node 57, containing 27 training observations, with node mean 0.259 and weight 0.00299 :
## 0.71 <= SER_countor
## texture_sumentropy_nondir_post4 <= 2
## 140 <= texture_variance_nondir_post2
## irregularity <= 0.97
## 1.8 <= texture_sumentropy_nondir_post3
##
## ============ bagging trees treedata_all
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7745546 0.6605714 0.7306423
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.8185224 0.6994286 0.7696579
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8544407 0.7394286 0.820078
## maxinter: 5 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 4 5 25 0.8894813 0.6731429 0.7864646
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 1 20 0.7745546 0.6605714 0.7306423
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 2 20 0.8137215 0.688 0.7645558
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 7 3 20 0.8620513 0.6988571 0.80012
## maxinter: 5 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 8 5 20 0.9129683 0.7388571 0.8365846
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 1 15 0.7745546 0.6605714 0.7306423
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 10 2 15 0.8179395 0.6994286 0.7818127
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 11 3 15 0.8572505 0.7188571 0.7936675
## maxinter: 5 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 12 5 15 0.8904244 0.696 0.80012
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 10 0.7745546 0.6605714 0.7306423
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 10 0.8170684 0.6822857 0.7677071
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 10 0.8719544 0.7222857 0.8181273
## maxinter: 5 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 16 5 10 0.8975177 0.7142857 0.8241297
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 17 1 5 0.7745546 0.6605714 0.7306423
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 18 2 5 0.8228321 0.6805714 0.7668067
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 19 3 5 0.8486704 0.7022857 0.77506
## maxinter: 5 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 20 5 5 0.8996987 0.7137143 0.807473
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7745546 0.6605714 0.7306423
## 2 2 25 0.8185224 0.6994286 0.7696579
## 3 3 25 0.8544407 0.7394286 0.8200780
## 4 5 25 0.8894813 0.6731429 0.7864646
## 5 1 20 0.7745546 0.6605714 0.7306423
## 6 2 20 0.8137215 0.6880000 0.7645558
## 7 3 20 0.8620513 0.6988571 0.8001200
## 8 5 20 0.9129683 0.7388571 0.8365846
## 9 1 15 0.7745546 0.6605714 0.7306423
## 10 2 15 0.8179395 0.6994286 0.7818127
## 11 3 15 0.8572505 0.7188571 0.7936675
## 12 5 15 0.8904244 0.6960000 0.8001200
## 13 1 10 0.7745546 0.6605714 0.7306423
## 14 2 10 0.8170684 0.6822857 0.7677071
## 15 3 10 0.8719544 0.7222857 0.8181273
## 16 5 10 0.8975177 0.7142857 0.8241297
## 17 1 5 0.7745546 0.6605714 0.7306423
## 18 2 5 0.8228321 0.6805714 0.7668067
## 19 3 5 0.8486704 0.7022857 0.7750600
## 20 5 5 0.8996987 0.7137143 0.8074730
## maxinter nodesize rocTrain rocValid rocTest
## 8 5 20 0.9129683 0.7388571 0.8365846
##
## Observation 1 has a predicted value 0.522
## since this is the weighted average response across the 7 nodes it is a member of:
##
## 1) Node 46, containing 36 training observations, with node mean 0.444 and weight 0.25 :
## peakCr_inside <= 3.5
## 6.9 <= V14
## T2texture_contrast_nondir <= 200
##
## 2) Node 62, containing 71 training observations, with node mean 0.333 and weight 0.226 :
## texture_variance_nondir_post1 <= 190
## 0.83 <= earlySE12
## 1.1 <= lateSE0
## 0.14 <= Vr_post_1_countor
## 1.2 <= dce3SE18
##
## 3) Node 61, containing 71 training observations, with node mean 0.901 and weight 0.21 :
## 1.2 <= dce3SE16
## 12 <= V18
## 160 <= texture_variance_nondir_post1
##
## 4) Node 70, containing 142 training observations, with node mean 0.338 and weight 0.12 :
## texture_variance_nondir_post1 <= 190
## 0.14 <= Vr_post_1_countor
## Vr_increasingRate_inside <= 1.1
## V12 <= 28
##
## 5) Node 43, containing 31 training observations, with node mean 0.806 and weight 0.0925 :
## iiMin_change_Variance_uptake <= 0.4
## V13 <= 13
## 0.74 <= UptakeRate_inside
## 4.9 <= V17
##
## 6) Node 68, containing 139 training observations, with node mean 0.302 and weight 0.092 :
## texture_variance_nondir_post1 <= 190
## max_F_r_i <= 1700
## 0.14 <= Vr_post_1_countor
## V18 <= 34
## 0.1 <= Vr_increasingRate_inside
##
## 7) Node 74, containing 567 training observations, with node mean 0.388 and weight 0.01 :
## ROOT NODE
## id C NC pred obs
## 1 1 0.4580405 0.5419595 NC C
## 2 32 0.3471798 0.6528202 NC NC
## 3 40 0.5146430 0.4853570 C C
## 4 41 0.3650571 0.6349429 NC C
## 5 51 0.3115149 0.6884851 NC NC
## 6 103 0.4246343 0.5753657 NC NC
## id C NC pred obs
## 1 1 0.5181722 0.4818278 C C
## 2 32 0.4096283 0.5903717 NC NC
## 3 40 0.4528250 0.5471750 NC C
## 4 41 0.3905985 0.6094015 NC C
## 5 51 0.3764796 0.6235204 NC NC
## 6 103 0.4017436 0.5982564 NC NC
## id C NC pred obs
## 1 1 0.4265003 0.5734997 NC C
## 2 32 0.3467061 0.6532939 NC NC
## 3 40 0.3257865 0.6742135 NC C
## 4 41 0.3382852 0.6617148 NC C
## 5 51 0.2661202 0.7338798 NC NC
## 6 103 0.3724694 0.6275306 NC NC
## id C NC pred obs
## 1 1 0.5223465 0.4776535 C C
## 2 32 0.2732180 0.7267820 NC NC
## 3 40 0.5411206 0.4588794 C C
## 4 41 0.3876384 0.6123616 NC C
## 5 51 0.2455812 0.7544188 NC NC
## 6 103 0.1250101 0.8749899 NC NC
##
## Call:
## roc.default(response = perf_imgT2$obs, predictor = perf_imgT2$C)
##
## Data: perf_imgT2$C in 48 controls (perf_imgT2$obs C) > 69 cases (perf_imgT2$obs NC).
## Area under the curve: 0.8
##
## Call:
## roc.default(response = perf_allT2$obs, predictor = perf_allT2$C)
##
## Data: perf_allT2$C in 47 controls (perf_allT2$obs C) > 70 cases (perf_allT2$obs NC).
## Area under the curve: 0.8012
##
## Call:
## roc.default(response = perf_imgT1$obs, predictor = perf_imgT1$C)
##
## Data: perf_imgT1$C in 55 controls (perf_imgT1$obs C) > 62 cases (perf_imgT1$obs NC).
## Area under the curve: 0.8399
##
## Call:
## roc.default(response = perf_all$obs, predictor = perf_all$C)
##
## Data: perf_all$C in 49 controls (perf_all$obs C) > 68 cases (perf_all$obs NC).
## Area under the curve: 0.8366
## Area under the curve: 0.8
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.409241 0.4792 0.625 0.7505 0.7536 0.8406 0.9275
## Area under the curve: 0.8012
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.3597217 0.8298 0.9149 0.9787 0.4429 0.5714 0.6857
## Area under the curve: 0.8399
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.3245263 0.8182 0.9091 0.9818 0.5968 0.7097 0.8226
## Area under the curve: 0.8366
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.4234976 0.6122 0.7347 0.8571 0.75 0.8382 0.9265
## massB massM nonmassB nonmassM
## 210 151 132 63
## massB massM nonmassB nonmassM
## 30 17 10 14
## massB massM nonmassB nonmassM
## 210 151 132 63
## massB massM nonmassB nonmassM
## 30 17 10 14
## massB massM nonmassB nonmassM
## 210 151 132 63
## massB massM nonmassB nonmassM
## 30 17 10 14
## 0.05392157 0.05
## 0.01036269 0.05
## Selected features for group: MeanDecreaseGini imgT2
## =========NULL
## [1] "T2texture_entropy_nondir" "ave_T210"
## [3] "T2texture_inversediffmoment_nondir" "T2RGH_mean"
## [5] "T2var_F_r_i" "ave_T27"
## [7] "T2texture_correlation_nondir" "ave_T213"
## [9] "T2skew_F_r_i" "ave_T211"
## [11] "T2texture_sumaverage_nondir" "T2max_F_r_i"
## [13] "T2RGH_var" "T2texture_diffentropy_nondir"
## [15] "ave_T24" "ave_T20"
## [17] "ave_T22" "ave_T216"
## [19] "ave_T219" "T2texture_energy_nondir"
## [21] "T2mean_F_r_i" "ave_T26"
## [23] "T2texture_variance_nondir" "T2grad_margin_var"
## [25] "ave_T215" "ave_T21"
## [27] "T2kurt_F_r_i" "ave_T29"
## [29] "ave_T218" "ave_T25"
## [31] "T2min_F_r_i" "ave_T23"
## [33] "ave_T217" "ave_T212"
## [35] "ave_T214" "ave_T28"
## [37] "T2grad_margin" "T2texture_sumvariance_nondir"
## [39] "T2_lesionSI"
## 0.01408451 0.05
## Selected features for group: MeanDecreaseGini allT2
## =========NULL
## [1] "T2RGH_mean" "T2texture_correlation_nondir"
## [3] "T2RGH_var" "T2max_F_r_i"
## [5] "T2kurt_F_r_i" "ave_T27"
## [7] "T2grad_margin_var" "T2texture_inversediffmoment_nondir"
## [9] "T2texture_sumaverage_nondir" "T2texture_energy_nondir"
## [11] "LMSIR_predicted" "ave_T210"
## [13] "ave_T214" "ave_T21"
## [15] "ave_T20" "ave_T212"
## [17] "ave_T215" "T2wSI_predicted"
## [19] "ave_T219" "T2texture_sumvariance_nondir"
## [21] "ave_T26" "ave_T28"
## [23] "T2texture_diffentropy_nondir" "ave_T217"
## [25] "ave_T29" "ave_T211"
## [27] "ave_T25" "T2texture_contrast_nondir"
## [29] "T2_lesionSI" "ave_T216"
## [31] "T2texture_sumentropy_nondir" "ave_T24"
## [33] "T2skew_F_r_i" "T2min_F_r_i"
## [35] "T2texture_diffvariance_nondir" "ave_T213"
## [37] "ave_T22" "T2_lesionSIstd"
## -0.07741935 0.05
## Selected features for group: MeanDecreaseGini imgT1
## =========NULL
## [1] "irregularity" "SER_inside"
## [3] "texture_diffvariance_nondir_post2" "texture_variance_nondir_post2"
## [5] "texture_sumaverage_nondir_post3" "texture_correlation_nondir_post3"
## [7] "var_F_r_i" "Kpeak_inside"
## [9] "texture_diffentropy_nondir_post4" "texture_entropy_nondir_post3"
## [11] "earlySE0" "V4"
## [13] "V0" "dce2SE5"
## [15] "dce2SE3" "V7"
## [17] "dce3SE4" "dce2SE13"
## [19] "lateSE6" "Vr_increasingRate_inside"
## [21] "earlySE10" "maxVr_countor"
## [23] "SER_countor" "iMax_Variance_uptake"
## [25] "dce2SE14" "dce2SE17"
## [27] "Kpeak_countor" "UptakeRate_inside"
## [29] "A_inside"
## 0.05128205 0.05
## 0.07432432 0.05
## -0.08029197 0.05
## Selected features for group: MeanDecreaseGini all
## =========NULL
## [1] "texture_sumvariance_nondir_post1" "texture_inversediffmoment_nondir_post2"
## [3] "V11" "iiMin_change_Variance_uptake"
## [5] "V6" "texture_diffvariance_nondir_post3"
## [7] "beta_inside" "max_RGH_mean"
## [9] "V16" "earlySE12"
## [11] "maxVr_inside" "dce2SE8"
## [13] "lateSE17" "T2kurt_F_r_i"
## [15] "kurt_F_r_i" "V13"
## [17] "ave_T212" "iiiMax_Margin_Gradient"
## [19] "UptakeRate_countor" "lateSE1"
## [21] "dce2SE13" "dce3SE0"
## [23] "lateSE2" "T2_lesionSIstd"
## [25] "ave_T29" "ave_T217"
## [27] "dce2SE2" "V3"
## [29] "Slope_ini_countor" "var_F_r_i"
## [31] "V15" "dce2SE6"
## [33] "Tpeak_inside" "T2texture_contrast_nondir"
## [35] "lateSE7" "A_inside"
## lesion_id cad_pt_no_txt exam_a_number_txt BIRADS lesion_label
## 23 23 0132 5154279 3 massB
## 24 24 0132 5154279 3 massB
## 54 54 0220 6715021 5 massB
## 55 55 0220 6715021 5 massB
## 57 57 0232 6671713 5 nonmassB
## 58 58 0232 6671713 5 nonmassB
## 66 66 0266 5254958 4 nonmassB
## 74 74 0325 4696948 4 massB
## 87 87 0442 4936886 4 massB
## 89 89 0462 5466989 3 nonmassM
## 90 90 0462 5466989 4 massB
## 102 102 0552 4663314 4 massB
## 112 112 0580 6855384 4 massB
## 123 123 0624 4894714 5 massB
## 139 139 0683 5226149 5 massM
## 172 172 0726 5304228 5 massM
## 188 188 0743 4827839 4 massM
## 209 209 0781 4738440 5 nonmassM
## 235 235 0812 4700538 5 massM
## 242 242 0818 5021762 4 massB
## 250 250 0831 4633368 6 nonmassM
## 282 282 0867 5372277 5 nonmassM
## 287 287 0873 4956191 4 massB
## 288 288 0873 4956191 4 massB
## 292 292 0877 4724338 4 nonmassB
## 293 293 0880 4809515 4 massB
## 294 294 0880 6778829 3 massM
## 295 295 0880 6778829 3 nonmassM
## 296 296 0880 6778829 3 nonmassM
## 297 297 0880 4809515 4 massB
## 299 299 0884 6876318 6 massM
## 300 300 0884 6876318 6 nonmassM
## 322 322 0952 7105222 4 nonmassB
## 323 323 0952 7105222 4 nonmassB
## 327 327 0965 6676125 3 massB
## 359 359 1071 7382882 4 nonmassM
## 360 360 1072 7554174 6 massM
## 361 361 1072 7554174 4 massB
## 373 373 1095 4378323 3 massB
## 374 374 1095 4378323 5 nonmassM
## 383 383 2028 6702914 6 nonmassB
## 384 384 2028 6702914 6 massM
## 385 385 2029 6716423 6 massB
## 398 398 2055 7041426 6 nonmassM
## 411 411 3004 7691918 4 nonmassB
## 412 412 3004 7691918 4 nonmassB
## 435 435 3046 7682447 4 massB
## 436 436 3046 7289130 4 massB
## 437 437 3046 7682447 4 massB
## 457 457 3075 7064471 6 massM
## 458 458 3075 7064471 6 nonmassM
## 474 474 3097 6909883 4 massB
## 475 475 4002 6993690 5 massB
## 487 487 4023 7037125 4 massM
## 488 488 4023 7037125 4 massB
## 489 489 4023 7152678 4 massB
## 490 490 4023 7037125 4 massM
## 506 506 6001 4574766 6 nonmassM
## 513 513 6014 5101372 6 massM
## 570 570 6052 5369136 6 nonmassM
## 571 571 6054 5425486 5 massM
## 572 572 6054 5425486 5 massM
## 573 573 6054 5425486 5 massM
## 574 574 6054 5425486 5 massM
## 575 575 6054 5425486 5 massM
## 592 592 6233 7047121 6 massB
## 604 604 7066 6715383 4 massB
## 605 605 7066 7395276 4 nonmassB
## 615 615 7096 6869668 3 massB
## 616 616 7096 6869668 3 massB
## 624 624 7159 5435020 4 nonmassM
## lesion_diagnosis find_t2_signal_int
## 23 BENIGN BREAST TISSUE Hyperintense
## 24 BENIGN BREAST TISSUE Hyperintense
## 54 RadialScar Slightly hyperintense
## 55 FIBROADENOMA Slightly hyperintense
## 57 FIBROCYSTIC None
## 58 FIBROCYSTIC None
## 66 FIBROCYSTIC Hypointense or not seen
## 74 FIBROCYSTIC Hyperintense
## 87 BENIGN BREAST TISSUE Slightly hyperintense
## 89 InvasiveDuctal Hypointense or not seen
## 90 FIBROADENOMA Hyperintense
## 102 ADENOSIS Slightly hyperintense
## 112 FIBROADENOMA None
## 123 FIBROCYSTIC Hypointense or not seen
## 139 InvasiveDuctal None
## 172 InvasiveDuctal Slightly hyperintense
## 188 InvasiveDuctal Hypointense or not seen
## 209 InvasiveDuctal None
## 235 InvasiveDuctal Hypointense or not seen
## 242 DUCT PAPILLOMA None
## 250 InsituDuctal Slightly hyperintense
## 282 InvasiveDuctal None
## 287 FIBROCYSTIC Hypointense or not seen
## 288 FIBROCYSTIC Slightly hyperintense
## 292 ATYPICAL LOBULAR HYPERPLASIA None
## 293 Papillary(focalAtypia) Slightly hyperintense
## 294 InsituDuctal None
## 295 InsituDuctal None
## 296 InsituDuctal None
## 297 AtypicalPapilloma Slightly hyperintense
## 299 InvasiveDuctal None
## 300 InsituDuctal None
## 322 FOCAL USUAL DUCTAL HYPERPLASIA Hypointense or not seen
## 323 FIBROADENOMA Hyperintense
## 327 FIBROADENOMA Slightly hyperintense
## 359 InsituDuctal None
## 360 InvasiveDuctal Slightly hyperintense
## 361 SCLEROSING PAPILLARY LESION Hypointense or not seen
## 373 FIBROADENOMA Hyperintense
## 374 InsituDuctal None
## 383 FIBROCYSTIC Hypointense or not seen
## 384 InvasiveDuctal None
## 385 ADENOSIS None
## 398 InvasiveDuctal None
## 411 FIBROCYSTIC None
## 412 FIBROCYSTIC None
## 435 FIBROADENOMA Hypointense or not seen
## 436 FIBROADENOMA Hyperintense
## 437 FIBROADENOMA Hypointense or not seen
## 457 InvasiveDuctal None
## 458 InsituDuctal None
## 474 ATYPICAL DUCTAL HYPERPLASIA Hypointense or not seen
## 475 BENIGN BREAST TISSUE Hypointense or not seen
## 487 InvasiveDuctal None
## 488 ADENOSIS, COLUMNAR CELL CHANGES Hypointense or not seen
## 489 BENIGN INTRAMAMMARY LYMPH NODE Hypointense or not seen
## 490 InvasiveDuctal None
## 506 InvasiveDuctal None
## 513 InvasiveDuctal None
## 570 InvasiveDuctal None
## 571 InsituDuctal None
## 572 InsituDuctal None
## 573 InsituDuctal None
## 574 InsituDuctal None
## 575 InsituDuctal None
## 592 FIBROADENOMA Hyperintense
## 604 FIBROADENOMA None
## 605 COLUMNAR CELL CHANGES None
## 615 FIBROADENOMA None
## 616 FIBROADENOMA Hypointense or not seen
## 624 InvasiveLobular None
##
## ============ bagging trees treedata_imgT2
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6688118 0.7685484 0.7176692
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.7551443 0.7616935 0.7685464
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8015522 0.6923387 0.7364662
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 4 1 20 0.6688118 0.7685484 0.7176692
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 2 20 0.74655 0.7197581 0.7233083
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 3 20 0.8230038 0.6975806 0.7317043
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 7 1 15 0.6688118 0.7685484 0.7176692
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 8 2 15 0.7553834 0.6705645 0.7037594
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 3 15 0.8358269 0.7125 0.7498747
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 10 1 10 0.6688118 0.7685484 0.7176692
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 11 2 10 0.7558821 0.7544355 0.7388471
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 12 3 10 0.8075368 0.6758065 0.7364662
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 5 0.6688118 0.7685484 0.7176692
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 5 0.7347516 0.6653226 0.7032581
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 5 0.8167527 0.6419355 0.7027569
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6688118 0.7685484 0.7176692
## 2 2 25 0.7551443 0.7616935 0.7685464
## 3 3 25 0.8015522 0.6923387 0.7364662
## 4 1 20 0.6688118 0.7685484 0.7176692
## 5 2 20 0.7465500 0.7197581 0.7233083
## 6 3 20 0.8230038 0.6975806 0.7317043
## 7 1 15 0.6688118 0.7685484 0.7176692
## 8 2 15 0.7553834 0.6705645 0.7037594
## 9 3 15 0.8358269 0.7125000 0.7498747
## 10 1 10 0.6688118 0.7685484 0.7176692
## 11 2 10 0.7558821 0.7544355 0.7388471
## 12 3 10 0.8075368 0.6758065 0.7364662
## 13 1 5 0.6688118 0.7685484 0.7176692
## 14 2 5 0.7347516 0.6653226 0.7032581
## 15 3 5 0.8167527 0.6419355 0.7027569
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.7551443 0.7616935 0.7685464
##
## Observation 1 has a predicted value 0.405
## since this is the weighted average response across the 8 nodes it is a member of:
##
## 1) Node 27, containing 383 training observations, with node mean 0.46 and weight 0.304 :
## T2min_F_r_i <= 9.5
## 56 <= ave_T210
##
## 2) Node 28, containing 384 training observations, with node mean 0.367 and weight 0.166 :
## T2texture_entropy_nondir <= 3.5
## T2texture_variance_nondir <= 440
##
## 3) Node 22, containing 262 training observations, with node mean 0.275 and weight 0.148 :
## T2texture_entropy_nondir <= 3.4
## T2texture_correlation_nondir <= 0.26
##
## 4) Node 25, containing 364 training observations, with node mean 0.363 and weight 0.145 :
## T2texture_entropy_nondir <= 3.4
## T2texture_variance_nondir <= 430
##
## 5) Node 23, containing 287 training observations, with node mean 0.488 and weight 0.131 :
## 350 <= T2RGH_var
## T2RGH_mean <= 54
##
## 6) Node 26, containing 381 training observations, with node mean 0.448 and weight 0.0852 :
## 3000 <= T2var_F_r_i
## 58 <= ave_T210
##
## 7) Node 24, containing 295 training observations, with node mean 0.485 and weight 0.0112 :
## 350 <= T2RGH_var
## 56 <= ave_T210
##
## 8) Node 29, containing 556 training observations, with node mean 0.385 and weight 0.01 :
## ROOT NODE
##
## ============ bagging trees treedata_allT2
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6760056 0.8306452 0.8
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.760637 0.7919355 0.7938961
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8120389 0.7024194 0.7437662
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 4 1 20 0.6760056 0.8306452 0.8
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 2 20 0.7429975 0.7483871 0.7768831
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 3 20 0.8075914 0.728629 0.7809091
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 7 1 15 0.6766068 0.833871 0.8032468
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 8 2 15 0.7458668 0.7681452 0.7816883
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 3 15 0.8028639 0.7548387 0.7832468
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 10 1 10 0.6760056 0.8306452 0.8
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 11 2 10 0.7556977 0.7645161 0.7848052
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 12 3 10 0.7919263 0.6895161 0.7706494
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 5 0.6766068 0.833871 0.8032468
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 5 0.7548369 0.7806452 0.787013
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 5 0.814225 0.7403226 0.7863636
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6760056 0.8306452 0.8000000
## 2 2 25 0.7606370 0.7919355 0.7938961
## 3 3 25 0.8120389 0.7024194 0.7437662
## 4 1 20 0.6760056 0.8306452 0.8000000
## 5 2 20 0.7429975 0.7483871 0.7768831
## 6 3 20 0.8075914 0.7286290 0.7809091
## 7 1 15 0.6766068 0.8338710 0.8032468
## 8 2 15 0.7458668 0.7681452 0.7816883
## 9 3 15 0.8028639 0.7548387 0.7832468
## 10 1 10 0.6760056 0.8306452 0.8000000
## 11 2 10 0.7556977 0.7645161 0.7848052
## 12 3 10 0.7919263 0.6895161 0.7706494
## 13 1 5 0.6766068 0.8338710 0.8032468
## 14 2 5 0.7548369 0.7806452 0.7870130
## 15 3 5 0.8142250 0.7403226 0.7863636
## maxinter nodesize rocTrain rocValid rocTest
## 7 1 15 0.6766068 0.833871 0.8032468
## 13 1 5 0.6766068 0.833871 0.8032468
##
## Observation 1 has a predicted value 0.371
## since this is the weighted average response across the 8 nodes it is a member of:
##
## 1) Node 8, containing 0.5 training observations, with node mean 0.296 and weight 0.348 :
## T2wSI_predicted in {Hyperintense,Hypointense or not seen,Slightly hyperintense}
##
## 2) Node 9, containing 318 training observations, with node mean 0.458 and weight 0.219 :
## 350 <= T2RGH_var
##
## 3) Node 14, containing 501 training observations, with node mean 0.41 and weight 0.13 :
## 56 <= ave_T210
##
## 4) Node 13, containing 450 training observations, with node mean 0.424 and weight 0.125 :
## 22 <= T2RGH_mean
##
## 5) Node 12, containing 437 training observations, with node mean 0.343 and weight 0.0947 :
## 0.00054 <= T2texture_energy_nondir
##
## 6) Node 10, containing 340 training observations, with node mean 0.326 and weight 0.068 :
## T2texture_correlation_nondir <= 0.26
##
## 7) Node 15, containing 553 training observations, with node mean 0.385 and weight 0.01 :
## ROOT NODE
##
## 8) Node 11, containing 430 training observations, with node mean 0.426 and weight 0.00523 :
## T2min_F_r_i <= 9.5
##
## ============ bagging trees treedata_imgT1
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.8130978 0.7379032 0.7528924
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.8328141 0.7495968 0.7624497
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8744603 0.8016129 0.8116197
## maxinter: 5 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 4 5 25 0.8992936 0.7370968 0.7882294
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 1 20 0.8139108 0.7395161 0.750503
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 2 20 0.8362983 0.7439516 0.7767857
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 7 3 20 0.8820367 0.7508065 0.7887324
## maxinter: 5 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 8 5 20 0.9013021 0.7379032 0.8106137
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 1 15 0.8139108 0.7387097 0.75
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 10 2 15 0.8396185 0.7580645 0.7794266
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 11 3 15 0.8677242 0.7790323 0.8202968
## maxinter: 5 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 12 5 15 0.9135582 0.7693548 0.8194165
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 10 0.8120594 0.7350806 0.7517606
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 10 0.8385186 0.7407258 0.7624497
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 10 0.87914 0.7697581 0.7999245
## maxinter: 5 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 16 5 10 0.9015822 0.7709677 0.8244467
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 17 1 5 0.8121618 0.7387097 0.7447183
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 18 2 5 0.8421667 0.7766129 0.7842052
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 19 3 5 0.8815243 0.7346774 0.7737676
## maxinter: 5 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 20 5 5 0.9023132 0.7987903 0.8279678
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.8130978 0.7379032 0.7528924
## 2 2 25 0.8328141 0.7495968 0.7624497
## 3 3 25 0.8744603 0.8016129 0.8116197
## 4 5 25 0.8992936 0.7370968 0.7882294
## 5 1 20 0.8139108 0.7395161 0.7505030
## 6 2 20 0.8362983 0.7439516 0.7767857
## 7 3 20 0.8820367 0.7508065 0.7887324
## 8 5 20 0.9013021 0.7379032 0.8106137
## 9 1 15 0.8139108 0.7387097 0.7500000
## 10 2 15 0.8396185 0.7580645 0.7794266
## 11 3 15 0.8677242 0.7790323 0.8202968
## 12 5 15 0.9135582 0.7693548 0.8194165
## 13 1 10 0.8120594 0.7350806 0.7517606
## 14 2 10 0.8385186 0.7407258 0.7624497
## 15 3 10 0.8791400 0.7697581 0.7999245
## 16 5 10 0.9015822 0.7709677 0.8244467
## 17 1 5 0.8121618 0.7387097 0.7447183
## 18 2 5 0.8421667 0.7766129 0.7842052
## 19 3 5 0.8815243 0.7346774 0.7737676
## 20 5 5 0.9023132 0.7987903 0.8279678
## maxinter nodesize rocTrain rocValid rocTest
## 20 5 5 0.9023132 0.7987903 0.8279678
##
## Observation 1 has a predicted value 0.461
## since this is the weighted average response across the 10 nodes it is a member of:
##
## 1) Node 70, containing 37 training observations, with node mean 0.459 and weight 0.27 :
## 0.71 <= SER_countor
## texture_correlation_nondir_post3 <= 0.37
## 0.92 <= irregularity
## -0.023 <= Kpeak_inside
##
## 2) Node 71, containing 37 training observations, with node mean 0.216 and weight 0.172 :
## SER_inside <= 0.8
## texture_variance_nondir_post2 <= 180
## 0.96 <= irregularity
## 3 <= texture_entropy_nondir_post3
##
## 3) Node 101, containing 79 training observations, with node mean 0.0759 and weight 0.136 :
## texture_variance_nondir_post2 <= 190
## SER_inside <= 0.8
## V7 <= 8.1
##
## 4) Node 37, containing 23 training observations, with node mean 0.957 and weight 0.119 :
## 0.33 <= UptakeRate_inside <= 0.67
## 0.93 <= irregularity
## A_inside <= 1.6
## 0.7 <= SER_countor
##
## 5) Node 91, containing 57 training observations, with node mean 0.263 and weight 0.0965 :
## 0.71 <= SER_countor
## texture_variance_nondir_post2 <= 310
## 3.7 <= V4
## 1.4 <= texture_diffentropy_nondir_post4
## texture_diffvariance_nondir_post2 <= 140
##
## 6) Node 105, containing 94 training observations, with node mean 0.83 and weight 0.0886 :
## 0.97 <= irregularity
## A_inside <= 4.3
## 0.67 <= SER_inside
##
## 7) Node 107, containing 100 training observations, with node mean 0.84 and weight 0.0607 :
## 0.72 <= SER_inside
## 0.97 <= irregularity
##
## 8) Node 49, containing 28 training observations, with node mean 0.464 and weight 0.041 :
## UptakeRate_inside <= 0.66
## 54000 <= var_F_r_i
## 0.91 <= lateSE6
## 0.35 <= Vr_increasingRate_inside
##
## 9) Node 119, containing 556 training observations, with node mean 0.385 and weight 0.01 :
## ROOT NODE
##
## 10) Node 89, containing 55 training observations, with node mean 0.491 and weight 0.00522 :
## texture_variance_nondir_post2 <= 180
## 0.96 <= irregularity
## 0.6 <= SER_countor
## 1.2 <= texture_diffentropy_nondir_post4
##
## ============ bagging trees treedata_all
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7743346 0.7028226 0.7067823
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.8167459 0.7431452 0.7692504
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.863188 0.7330645 0.7993371
## maxinter: 5 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 4 5 25 0.9133874 0.6943548 0.7901581
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 1 20 0.7743346 0.7028226 0.7067823
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 2 20 0.8275332 0.7556452 0.7608363
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 7 3 20 0.8666448 0.7758065 0.8215196
## maxinter: 5 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 8 5 20 0.9049093 0.7419355 0.8105558
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 1 15 0.7743346 0.7028226 0.7067823
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 10 2 15 0.8300404 0.7508065 0.7756247
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 11 3 15 0.8681683 0.7580645 0.8100459
## maxinter: 5 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 12 5 15 0.9084139 0.775 0.8355431
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 10 0.7743346 0.7028226 0.7067823
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 10 0.8369678 0.7737903 0.8081336
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 10 0.8669591 0.7556452 0.8021418
## maxinter: 5 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 16 5 10 0.9145762 0.7064516 0.8131056
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 17 1 5 0.7743346 0.7028226 0.7067823
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 18 2 5 0.8279568 0.7548387 0.781489
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 19 3 5 0.8626961 0.75 0.795895
## maxinter: 5 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 20 5 5 0.9124242 0.8056452 0.8623151
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7743346 0.7028226 0.7067823
## 2 2 25 0.8167459 0.7431452 0.7692504
## 3 3 25 0.8631880 0.7330645 0.7993371
## 4 5 25 0.9133874 0.6943548 0.7901581
## 5 1 20 0.7743346 0.7028226 0.7067823
## 6 2 20 0.8275332 0.7556452 0.7608363
## 7 3 20 0.8666448 0.7758065 0.8215196
## 8 5 20 0.9049093 0.7419355 0.8105558
## 9 1 15 0.7743346 0.7028226 0.7067823
## 10 2 15 0.8300404 0.7508065 0.7756247
## 11 3 15 0.8681683 0.7580645 0.8100459
## 12 5 15 0.9084139 0.7750000 0.8355431
## 13 1 10 0.7743346 0.7028226 0.7067823
## 14 2 10 0.8369678 0.7737903 0.8081336
## 15 3 10 0.8669591 0.7556452 0.8021418
## 16 5 10 0.9145762 0.7064516 0.8131056
## 17 1 5 0.7743346 0.7028226 0.7067823
## 18 2 5 0.8279568 0.7548387 0.7814890
## 19 3 5 0.8626961 0.7500000 0.7958950
## 20 5 5 0.9124242 0.8056452 0.8623151
## maxinter nodesize rocTrain rocValid rocTest
## 20 5 5 0.9124242 0.8056452 0.8623151
##
## Observation 1 has a predicted value 0.403
## since this is the weighted average response across the 11 nodes it is a member of:
##
## 1) Node 124, containing 101 training observations, with node mean 0.406 and weight 0.258 :
## 0.1 <= iiiMax_Margin_Gradient
## max_RGH_mean <= 0.59
## texture_sumvariance_nondir_post1 <= 470
## T2kurt_F_r_i <= 4.8
## V15 <= 35
##
## 2) Node 68, containing 23 training observations, with node mean 0.304 and weight 0.187 :
## Slope_ini_countor <= 0.86
## texture_sumvariance_nondir_post1 <= 430
## V11 <= 30
## texture_inversediffmoment_nondir_post2 <= 0.19
## 55000 <= var_F_r_i
##
## 3) Node 127, containing 127 training observations, with node mean 0.126 and weight 0.135 :
## 0.23 <= iiMin_change_Variance_uptake
## Slope_ini_countor <= 0.46
## V15 <= 44
##
## 4) Node 55, containing 20 training observations, with node mean 0.75 and weight 0.101 :
## dce2SE8 <= 1
## 54000 <= var_F_r_i
## 0.42 <= Slope_ini_countor
##
## 5) Node 38, containing 17 training observations, with node mean 0.529 and weight 0.0697 :
## 6 <= Tpeak_inside
## 13 <= V16
## 0.23 <= iiMin_change_Variance_uptake
## A_inside <= 1.1
##
## 6) Node 16, containing 13 training observations, with node mean 0.923 and weight 0.0556 :
## texture_sumvariance_nondir_post1 <= 480
## 0.82 <= earlySE12
## texture_diffvariance_nondir_post3 <= 85
## 23000 <= var_F_r_i
## texture_inversediffmoment_nondir_post2 <= 0.21
##
## 7) Node 99, containing 52 training observations, with node mean 0.462 and weight 0.0509 :
## 6 <= Tpeak_inside
## 140 <= texture_sumvariance_nondir_post1 <= 380
## 19 <= V6
##
## 8) Node 132, containing 154 training observations, with node mean 0.156 and weight 0.0497 :
## dce2SE13 <= 1.3
## V15 <= 41
## 6 <= Tpeak_inside
## 0.22 <= iiMin_change_Variance_uptake
##
## 9) Node 93, containing 43 training observations, with node mean 0.372 and weight 0.0427 :
## Slope_ini_countor <= 0.82
## 250 <= texture_sumvariance_nondir_post1 <= 490
## 0.53 <= max_RGH_mean
## T2texture_contrast_nondir <= 520
##
## 10) Node 134, containing 167 training observations, with node mean 0.24 and weight 0.0412 :
## dce2SE13 <= 0.83
##
## 11) Node 138, containing 556 training observations, with node mean 0.385 and weight 0.01 :
## ROOT NODE
## id C NC pred obs
## 1 23 0.4050504 0.5949496 NC NC
## 2 24 0.3293008 0.6706992 NC NC
## 3 54 0.3843178 0.6156822 NC NC
## 4 55 0.3635240 0.6364760 NC NC
## 5 57 0.3635240 0.6364760 NC NC
## 6 58 0.5416345 0.4583655 C NC
## id C NC pred obs
## 1 23 0.3705154 0.6294846 NC NC
## 2 24 0.3705154 0.6294846 NC NC
## 3 54 0.3326995 0.6673005 NC NC
## 4 55 0.3063445 0.6936555 NC NC
## 5 57 0.3946538 0.6053462 NC NC
## 6 58 0.4508855 0.5491145 NC NC
## id C NC pred obs
## 1 23 0.4611776 0.5388224 NC NC
## 2 24 0.2819303 0.7180697 NC NC
## 3 54 0.5388048 0.4611952 C NC
## 4 55 0.1570961 0.8429039 NC NC
## 5 57 0.6302258 0.3697742 C NC
## 6 58 0.3478923 0.6521077 NC NC
## id C NC pred obs
## 1 23 0.4032408 0.5967592 NC NC
## 2 24 0.5083533 0.4916467 C NC
## 3 54 0.6776382 0.3223618 C NC
## 4 55 0.3354721 0.6645279 NC NC
## 5 57 0.5134632 0.4865368 C NC
## 6 58 0.5795618 0.4204382 C NC
##
## Call:
## roc.default(response = perf_imgT2$obs, predictor = perf_imgT2$C)
##
## Data: perf_imgT2$C in 105 controls (perf_imgT2$obs C) > 139 cases (perf_imgT2$obs NC).
## Area under the curve: 0.7877
##
## Call:
## roc.default(response = perf_allT2$obs, predictor = perf_allT2$C)
##
## Data: perf_allT2$C in 97 controls (perf_allT2$obs C) > 147 cases (perf_allT2$obs NC).
## Area under the curve: 0.7979
##
## Call:
## roc.default(response = perf_imgT1$obs, predictor = perf_imgT1$C)
##
## Data: perf_imgT1$C in 111 controls (perf_imgT1$obs C) > 133 cases (perf_imgT1$obs NC).
## Area under the curve: 0.8309
##
## Call:
## roc.default(response = perf_all$obs, predictor = perf_all$C)
##
## Data: perf_all$C in 102 controls (perf_all$obs C) > 142 cases (perf_all$obs NC).
## Area under the curve: 0.8486
## Area under the curve: 0.7877
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.3642906 0.7905 0.8571 0.9238 0.5036 0.5899 0.6691
## Area under the curve: 0.7979
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.3689379 0.8351 0.8969 0.9485 0.4966 0.5782 0.6599
## Area under the curve: 0.8309
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.3774603 0.6937 0.7748 0.8468 0.7293 0.797 0.8647
## Area under the curve: 0.8486
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.3694122 0.7353 0.8137 0.8824 0.7042 0.7746 0.838
## massB massM nonmassB nonmassM
## 217 155 129 71
## massB massM nonmassB nonmassM
## 23 13 13 6
## massB massM nonmassB nonmassM
## 217 155 129 71
## massB massM nonmassB nonmassM
## 23 13 13 6
## massB massM nonmassB nonmassM
## 217 155 129 71
## massB massM nonmassB nonmassM
## 23 13 13 6
## -0.009803922 0.05
## Selected features for group: MeanDecreaseGini imgT2
## =========NULL
## [1] "T2RGH_mean" "T2texture_correlation_nondir"
## [3] "T2RGH_var" "T2skew_F_r_i"
## [5] "ave_T210" "ave_T211"
## [7] "T2max_F_r_i" "ave_T212"
## [9] "T2_lesionSIstd" "ave_T216"
## [11] "T2grad_margin_var" "T2texture_sumaverage_nondir"
## [13] "ave_T28" "T2texture_diffentropy_nondir"
## [15] "ave_T218" "ave_T213"
## [17] "ave_T29" "ave_T26"
## [19] "ave_T219" "T2mean_F_r_i"
## [21] "T2min_F_r_i" "T2grad_margin"
## [23] "T2texture_sumvariance_nondir" "ave_T21"
## [25] "ave_T20" "T2texture_inversediffmoment_nondir"
## [27] "T2texture_energy_nondir" "ave_T214"
## [29] "ave_T22" "T2_lesionSI"
## 0.04265403 0.05
## Selected features for group: MeanDecreaseGini allT2
## =========NULL
## [1] "T2RGH_var" "T2texture_correlation_nondir"
## [3] "T2wSI_predicted" "T2RGH_mean"
## [5] "T2texture_energy_nondir" "T2skew_F_r_i"
## [7] "T2grad_margin" "T2_lesionSIstd"
## [9] "ave_T212" "ave_T28"
## [11] "ave_T211" "ave_T218"
## [13] "ave_T29" "ave_T21"
## [15] "T2max_F_r_i" "ave_T210"
## [17] "ave_T215" "ave_T25"
## [19] "T2texture_inversediffmoment_nondir" "T2_lesionSI"
## [21] "ave_T214" "T2kurt_F_r_i"
## [23] "T2texture_diffentropy_nondir" "T2texture_sumvariance_nondir"
## [25] "ave_T216" "ave_T27"
## [27] "ave_T22" "LMSIR_predicted"
## [29] "ave_T20" "ave_T217"
## [31] "T2texture_diffvariance_nondir" "ave_T219"
## [33] "ave_T213" "ave_T23"
## [35] "T2texture_sumaverage_nondir" "T2min_F_r_i"
## [37] "T2mean_F_r_i"
## -0.03246753 0.05
## Selected features for group: MeanDecreaseGini imgT1
## =========NULL
## [1] "texture_variance_nondir_post1" "V12"
## [3] "var_F_r_i" "texture_energy_nondir_post4"
## [5] "texture_correlation_nondir_post3" "texture_variance_nondir_post2"
## [7] "beta_inside" "V1"
## [9] "iiiMax_Margin_Gradient" "dce3SE7"
## [11] "Kpeak_countor" "V14"
## [13] "earlySE2" "V3"
## [15] "dce3SE8" "earlySE9"
## [17] "A_countor" "texture_entropy_nondir_post3"
## [19] "earlySE11" "V17"
## [21] "lateSE19" "max_RGH_var"
## [23] "iAUC1_countor" "V5"
## [25] "texture_contrast_nondir_post2" "dce2SE15"
## [27] "lateSE14" "lateSE18"
## [29] "A_inside"
## 0 0.05
## Selected features for group: MeanDecreaseGini all
## =========NULL
## [1] "SER_countor" "texture_correlation_nondir_post1"
## [3] "V12" "max_RGH_var"
## [5] "Vr_post_1_countor" "texture_sumentropy_nondir_post3"
## [7] "T2RGH_var" "dce3SE9"
## [9] "T2max_F_r_i" "V3"
## [11] "ave_T215" "T2texture_diffentropy_nondir"
## [13] "texture_diffentropy_nondir_post2" "dce3SE13"
## [15] "texture_sumaverage_nondir_post2" "ave_T218"
## [17] "beta_inside" "iAUC1_inside"
## [19] "lateSE6" "dce3SE6"
## [21] "ave_T25" "dce3SE1"
## [23] "texture_inversediffmoment_nondir_post1" "earlySE7"
## [25] "V8" "lateSE12"
## [27] "ave_T219" "lateSE18"
## [29] "A_inside"
## lesion_id cad_pt_no_txt exam_a_number_txt BIRADS lesion_label
## 33 33 0177 6996979 3 massM
## 34 34 0177 6996979 3 nonmassM
## 52 52 0207 4982884 4 massB
## 83 83 0420 6738142 3 nonmassM
## 93 93 0473 7364625 4 massB
## 96 96 0510 7662547 4 nonmassB
## 97 97 0510 7662547 4 nonmassB
## 99 99 0519 4937737 4 massB
## 121 121 0619 7250777 5 massM
## 122 122 0619 7250777 5 nonmassM
## 124 124 0635 7092156 4 massM
## 127 127 0663 4804825 4 massB
## 133 133 0668 6989634 4 nonmassM
## 146 146 0690 5180451 3 nonmassB
## 147 147 0690 5180451 3 massB
## 148 148 0690 6681276 4 nonmassB
## 171 171 0724 5141876 5 massM
## 198 198 0760 4750742 5 nonmassM
## 212 212 0758 4796378 4 massB
## 213 213 0758 4796378 4 massB
## 219 219 0790 4708057 4 nonmassB
## 230 230 0803 5058195 5 massM
## 233 233 0809 5016014 4 massB
## 303 303 0888 6744887 5 massM
## 304 304 0896 6895982 4 massB
## 305 305 0898 5224531 4 massB
## 311 311 0921 6997232 4 massB
## 318 318 0944 7742881 4 massM
## 319 319 0944 7092128 4 nonmassB
## 325 325 0956 5062341 4 massB
## 326 326 0962 4755483 4 massB
## 333 333 0995 6816787 4 nonmassB
## 334 334 0995 6816787 3 massB
## 339 339 1006 4443563 4 nonmassB
## 368 368 1087 5360576 4 massB
## 369 369 1087 5360576 4 massB
## 404 404 2072 7256932 4 massM
## 407 407 2075 6985605 4 nonmassB
## 408 408 2075 6985605 4 massB
## 419 419 3011 6898308 4 nonmassB
## 461 461 3077 7042083 4 nonmassB
## 462 462 3077 7042083 4 massB
## 480 480 4012 7002008 4 massB
## 498 498 4043 7041465 6 nonmassM
## 509 509 6005 ACC108250 5 massM
## 510 510 6005 ACC108250 5 massM
## 511 511 6005 ACC108250 5 massM
## 552 552 6040 5075204 5 massM
## 578 578 6100 6722170 5 massM
## 586 586 6150 7128025 4 nonmassB
## 597 597 7024 6805356 4 massB
## 601 601 7043 7119983 4 massB
## 623 623 7151 7557684 2 massB
## 625 625 7165 5021830 3 massB
## 635 635 7201 5041620 4 nonmassB
## lesion_diagnosis find_t2_signal_int
## 33 InsituDuctal Slightly hyperintense
## 34 InsituDuctal None
## 52 FIBROSIS None
## 83 InsituDuctal Slightly hyperintense
## 93 FIBROCYSTIC None
## 96 COLUMNAR CELL CHANGES None
## 97 COLUMNAR CELL CHANGES None
## 99 FLAT EPITHELIAL ATYPIA None
## 121 InvasiveDuctal None
## 122 InvasiveDuctal Hyperintense
## 124 InvasiveDuctal Hypointense or not seen
## 127 FIBROADENOMA Hyperintense
## 133 InsituDuctal None
## 146 FIBROCYSTIC None
## 147 USUAL DUCTAL HYPERPLASIA Slightly hyperintense
## 148 COLUMNAR CELL CHANGES Hypointense or not seen
## 171 InvasiveDuctal Hyperintense
## 198 InsituDuctal None
## 212 FIBROCYSTIC Hyperintense
## 213 FIBROCYSTIC Hyperintense
## 219 DUCT PAPILLOMA None
## 230 InvasiveDuctal None
## 233 FIBROCYSTIC Hyperintense
## 303 InvasiveDuctal None
## 304 FIBROADENOMA Slightly hyperintense
## 305 DUCTAL HYPERPLASIA WITHOUT ATYPIA None
## 311 FIBROCYSTIC Hypointense or not seen
## 318 InsituDuctal Hypointense or not seen
## 319 BENIGN BREAST TISSUE None
## 325 FIBROADENOMA Slightly hyperintense
## 326 FIBROADENOMA Hyperintense
## 333 LARGE DUCT PAPILLOMA Hyperintense
## 334 DUCT PAPILLOMA Slightly hyperintense
## 339 FIBROCYSTIC Hypointense or not seen
## 368 GRANULOMATOUS LOBULAR MASTITIS None
## 369 GRANULOMATOUS LOBULAR MASTITIS None
## 404 InvasiveDuctal None
## 407 BENIGN BREAST TISSUE Hypointense or not seen
## 408 FIBROADENOMA Hypointense or not seen
## 419 FIBROCYSTIC None
## 461 FIBROCYSTIC None
## 462 FIBROCYSTIC Slightly hyperintense
## 480 FOCAL USUAL DUCTAL HYPERPLASIA Hypointense or not seen
## 498 InvasiveDuctal None
## 509 InvasiveLobular None
## 510 InvasiveLobular None
## 511 InvasiveLobular None
## 552 InvasiveDuctal None
## 578 InvasiveDuctal Hypointense or not seen
## 586 COLUMNAR CELL CHANGES None
## 597 FIBROADENOMA Hypointense or not seen
## 601 FIBROADENOMA Hyperintense
## 623 HYPERPLASIA None
## 625 ADENOSIS None
## 635 FIBROADENOMA None
##
## ============ bagging trees treedata_imgT2
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6809044 0.6016082 0.5876712
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.7507865 0.5694444 0.6251712
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.7874124 0.4393275 0.6488014
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 4 1 20 0.6773173 0.6067251 0.5885274
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 2 20 0.7355491 0.4817251 0.602226
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 3 20 0.7880646 0.5942982 0.6458904
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 7 1 15 0.6771126 0.6067251 0.5875
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 8 2 15 0.7520525 0.5467836 0.5986301
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 3 15 0.8103292 0.6403509 0.7092466
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 10 1 10 0.6771126 0.6067251 0.5875
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 11 2 10 0.7526792 0.5782164 0.6409247
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 12 3 10 0.8282968 0.6315789 0.7006849
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 5 0.6773173 0.6067251 0.5885274
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 5 0.7606335 0.4612573 0.5964041
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 5 0.7883076 0.6527778 0.6825342
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6809044 0.6016082 0.5876712
## 2 2 25 0.7507865 0.5694444 0.6251712
## 3 3 25 0.7874124 0.4393275 0.6488014
## 4 1 20 0.6773173 0.6067251 0.5885274
## 5 2 20 0.7355491 0.4817251 0.6022260
## 6 3 20 0.7880646 0.5942982 0.6458904
## 7 1 15 0.6771126 0.6067251 0.5875000
## 8 2 15 0.7520525 0.5467836 0.5986301
## 9 3 15 0.8103292 0.6403509 0.7092466
## 10 1 10 0.6771126 0.6067251 0.5875000
## 11 2 10 0.7526792 0.5782164 0.6409247
## 12 3 10 0.8282968 0.6315789 0.7006849
## 13 1 5 0.6773173 0.6067251 0.5885274
## 14 2 5 0.7606335 0.4612573 0.5964041
## 15 3 5 0.7883076 0.6527778 0.6825342
## maxinter nodesize rocTrain rocValid rocTest
## 9 3 15 0.8103292 0.6403509 0.7092466
##
## Observation 1 has a predicted value 0.331
## since this is the weighted average response across the 9 nodes it is a member of:
##
## 1) Node 36, containing 88 training observations, with node mean 0.148 and weight 0.288 :
## T2_lesionSIstd <= 87
## 8.5 <= T2min_F_r_i
##
## 2) Node 53, containing 374 training observations, with node mean 0.402 and weight 0.199 :
## 0.00048 <= T2texture_energy_nondir
## T2texture_diffentropy_nondir <= 1.8
## 56 <= ave_T210
##
## 3) Node 51, containing 216 training observations, with node mean 0.35 and weight 0.157 :
## 0.00048 <= T2texture_energy_nondir
## 45 <= T2texture_sumaverage_nondir
## 99 <= ave_T28
##
## 4) Node 50, containing 208 training observations, with node mean 0.458 and weight 0.146 :
## 350 <= T2RGH_var
## 120 <= ave_T211
##
## 5) Node 52, containing 246 training observations, with node mean 0.392 and weight 0.0937 :
## 22 <= T2RGH_mean
## 41 <= ave_T210
## T2texture_correlation_nondir <= 0.26
##
## 6) Node 43, containing 159 training observations, with node mean 0.562 and weight 0.0764 :
## 350 <= T2RGH_var
## 54 <= ave_T210
## T2grad_margin_var <= 3300
##
## 7) Node 32, containing 71 training observations, with node mean 0.108 and weight 0.025 :
## 8.5 <= T2min_F_r_i
## ave_T213 <= 200
##
## 8) Node 54, containing 568 training observations, with node mean 0.395 and weight 0.01 :
## ROOT NODE
##
## 9) Node 34, containing 82 training observations, with node mean 0.134 and weight 0.00511 :
## T2texture_correlation_nondir <= 0.26
## T2max_F_r_i <= 710
## 6.5 <= T2min_F_r_i
##
## ============ bagging trees treedata_allT2
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6822855 0.6154971 0.6589474
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.758191 0.6630117 0.7457895
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.818194 0.5577485 0.7321053
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 4 1 20 0.6822855 0.6154971 0.6589474
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 2 20 0.7591501 0.6140351 0.7257895
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 3 20 0.8107384 0.6081871 0.7447368
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 7 1 15 0.6822855 0.6154971 0.6589474
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 8 2 15 0.7566947 0.6147661 0.7208772
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 3 15 0.8147603 0.5906433 0.7512281
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 10 1 10 0.6822855 0.6154971 0.6589474
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 11 2 10 0.7516881 0.6125731 0.7024561
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 12 3 10 0.801486 0.5972222 0.7385965
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 5 0.6822855 0.6154971 0.6589474
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 5 0.750665 0.624269 0.7235088
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 5 0.8313341 0.6147661 0.745614
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6822855 0.6154971 0.6589474
## 2 2 25 0.7581910 0.6630117 0.7457895
## 3 3 25 0.8181940 0.5577485 0.7321053
## 4 1 20 0.6822855 0.6154971 0.6589474
## 5 2 20 0.7591501 0.6140351 0.7257895
## 6 3 20 0.8107384 0.6081871 0.7447368
## 7 1 15 0.6822855 0.6154971 0.6589474
## 8 2 15 0.7566947 0.6147661 0.7208772
## 9 3 15 0.8147603 0.5906433 0.7512281
## 10 1 10 0.6822855 0.6154971 0.6589474
## 11 2 10 0.7516881 0.6125731 0.7024561
## 12 3 10 0.8014860 0.5972222 0.7385965
## 13 1 5 0.6822855 0.6154971 0.6589474
## 14 2 5 0.7506650 0.6242690 0.7235088
## 15 3 5 0.8313341 0.6147661 0.7456140
## maxinter nodesize rocTrain rocValid rocTest
## 9 3 15 0.8147603 0.5906433 0.7512281
##
## Observation 1 has a predicted value 0.371
## since this is the weighted average response across the 11 nodes it is a member of:
##
## 1) Node 68, containing 0.5 training observations, with node mean 0.306 and weight 0.316 :
## T2wSI_predicted in {Hyperintense,Hypointense or not seen,Slightly hyperintense}
## T2texture_inversediffmoment_nondir <= 0.25
## 0.1 <= T2texture_correlation_nondir
##
## 2) Node 71, containing 271 training observations, with node mean 0.372 and weight 0.195 :
## 22 <= T2RGH_mean
## T2texture_correlation_nondir <= 0.26
##
## 3) Node 64, containing 0.5 training observations, with node mean 0.617 and weight 0.144 :
## T2wSI_predicted in {None,Slightly hyperintense}
## 410 <= T2max_F_r_i
## T2kurt_F_r_i <= 4.6
##
## 4) Node 69, containing 0.5 training observations, with node mean 0.237 and weight 0.135 :
## T2wSI_predicted in {Hyperintense,Hypointense or not seen,Slightly hyperintense}
## ave_T213 <= 250
## 0.00044 <= T2texture_energy_nondir
##
## 5) Node 65, containing 0.5 training observations, with node mean 0.403 and weight 0.06 :
## T2wSI_predicted in {Hyperintense,Hypointense or not seen,Slightly hyperintense}
## 22 <= T2RGH_mean
## 0.12 <= T2texture_correlation_nondir
##
## 6) Node 72, containing 340 training observations, with node mean 0.438 and weight 0.0536 :
## 22 <= T2RGH_mean
## 0.00041 <= T2texture_energy_nondir
## 56 <= ave_T210
##
## 7) Node 40, containing 62 training observations, with node mean 0.177 and weight 0.0293 :
## 1.5 <= T2texture_diffentropy_nondir
## T2texture_correlation_nondir <= 0.15
## T2_lesionSIstd <= 87
##
## 8) Node 66, containing 0.5 training observations, with node mean 0.58 and weight 0.0281 :
## T2wSI_predicted in {None,Slightly hyperintense}
## 350 <= T2RGH_var
## 35 <= ave_T25
##
## 9) Node 54, containing 127 training observations, with node mean 0.236 and weight 0.0156 :
## T2texture_correlation_nondir <= 0.14
## ave_T214 <= 250
##
## 10) Node 51, containing 97 training observations, with node mean 0.299 and weight 0.0103 :
## T2texture_correlation_nondir <= 0.26
## 0.00034 <= T2texture_energy_nondir
## 650 <= T2texture_sumvariance_nondir
##
## 11) Node 73, containing 568 training observations, with node mean 0.395 and weight 0.01 :
## ROOT NODE
##
## ============ bagging trees treedata_imgT1
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7402872 0.625 0.6594771
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.8041588 0.5869883 0.7230392
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8533109 0.6593567 0.7823529
## maxinter: 5 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 4 5 25 0.8956277 0.3991228 0.7656863
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 1 20 0.7409139 0.6140351 0.645098
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 2 20 0.8143831 0.5716374 0.7037582
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 7 3 20 0.8554849 0.5811404 0.725817
## maxinter: 5 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 8 5 20 0.900078 0.6754386 0.7952614
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 1 15 0.7332984 0.5994152 0.627451
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 10 2 15 0.8117807 0.4817251 0.6844771
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 11 3 15 0.8578892 0.5964912 0.7334967
## maxinter: 5 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 12 5 15 0.8927055 0.6798246 0.7911765
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 10 0.7378191 0.5906433 0.6493464
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 10 0.8245562 0.5869883 0.7197712
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 10 0.8373638 0.5336257 0.700817
## maxinter: 5 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 16 5 10 0.9021433 0.627924 0.7707516
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 17 1 5 0.7409266 0.6125731 0.6511438
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 18 2 5 0.8113203 0.5950292 0.7114379
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 19 3 5 0.84551 0.5657895 0.7215686
## maxinter: 5 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 20 5 5 0.8877436 0.6571637 0.7730392
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7402872 0.6250000 0.6594771
## 2 2 25 0.8041588 0.5869883 0.7230392
## 3 3 25 0.8533109 0.6593567 0.7823529
## 4 5 25 0.8956277 0.3991228 0.7656863
## 5 1 20 0.7409139 0.6140351 0.6450980
## 6 2 20 0.8143831 0.5716374 0.7037582
## 7 3 20 0.8554849 0.5811404 0.7258170
## 8 5 20 0.9000780 0.6754386 0.7952614
## 9 1 15 0.7332984 0.5994152 0.6274510
## 10 2 15 0.8117807 0.4817251 0.6844771
## 11 3 15 0.8578892 0.5964912 0.7334967
## 12 5 15 0.8927055 0.6798246 0.7911765
## 13 1 10 0.7378191 0.5906433 0.6493464
## 14 2 10 0.8245562 0.5869883 0.7197712
## 15 3 10 0.8373638 0.5336257 0.7008170
## 16 5 10 0.9021433 0.6279240 0.7707516
## 17 1 5 0.7409266 0.6125731 0.6511438
## 18 2 5 0.8113203 0.5950292 0.7114379
## 19 3 5 0.8455100 0.5657895 0.7215686
## 20 5 5 0.8877436 0.6571637 0.7730392
## maxinter nodesize rocTrain rocValid rocTest
## 8 5 20 0.900078 0.6754386 0.7952614
##
## Observation 1 has a predicted value 0.179
## since this is the weighted average response across the 11 nodes it is a member of:
##
## 1) Node 112, containing 165 training observations, with node mean 0.187 and weight 0.265 :
## dce3SE7 <= 1.6
## V12 <= 20
## 5.5 <= V17
## texture_variance_nondir_post1 <= 210
## var_F_r_i <= 74000
##
## 2) Node 117, containing 212 training observations, with node mean 0.184 and weight 0.229 :
## texture_entropy_nondir_post3 <= 3.1
## texture_variance_nondir_post1 <= 110
## texture_energy_nondir_post4 <= 0.0056
## V12 <= 37
##
## 3) Node 116, containing 207 training observations, with node mean 0.164 and weight 0.16 :
## var_F_r_i <= 62000
## V12 <= 37
## texture_contrast_nondir_post2 <= 280
## texture_correlation_nondir_post3 <= 0.43
##
## 4) Node 82, containing 47 training observations, with node mean 0.149 and weight 0.0755 :
## texture_variance_nondir_post1 <= 170
## var_F_r_i <= 73000
## V1 <= 20
## lateSE18 <= 1
## lateSE19 <= 0.97
##
## 5) Node 77, containing 43 training observations, with node mean 0.209 and weight 0.072 :
## dce2SE15 <= 0.88
## A_inside <= 1.6
## texture_variance_nondir_post1 <= 67
##
## 6) Node 80, containing 45 training observations, with node mean 0.0667 and weight 0.0633 :
## texture_variance_nondir_post1 <= 78
## A_inside <= 1.4
## earlySE2 <= 0.74
## var_F_r_i <= 61000
##
## 7) Node 95, containing 70 training observations, with node mean 0.314 and weight 0.0467 :
## texture_variance_nondir_post1 <= 210
## texture_variance_nondir_post2 <= 300
## earlySE2 <= 0.74
## beta_inside <= 0.11
## dce2SE15 <= 0.7
##
## 8) Node 105, containing 106 training observations, with node mean 0.113 and weight 0.0395 :
## texture_variance_nondir_post1 <= 190
## V12 <= 38
## V14 <= 13
## texture_energy_nondir_post4 <= 0.0021
##
## 9) Node 115, containing 192 training observations, with node mean 0.193 and weight 0.0341 :
## texture_variance_nondir_post2 <= 300
## texture_variance_nondir_post1 <= 210
## 0.071 <= max_RGH_var
## var_F_r_i <= 74000
## earlySE2 <= 0.84
##
## 10) Node 118, containing 572 training observations, with node mean 0.395 and weight 0.01 :
## ROOT NODE
##
## 11) Node 114, containing 183 training observations, with node mean 0.205 and weight 0.00536 :
## dce3SE7 <= 1.6
## var_F_r_i <= 64000
## earlySE9 <= 1.1
## 1.2 <= A_countor
## texture_contrast_nondir_post2 <= 320
##
## ============ bagging trees treedata_all
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.749367 0.751462 0.7940039
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.8187056 0.8347953 0.8568665
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8596156 0.8011696 0.8596067
## maxinter: 5 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 4 5 25 0.8878267 0.745614 0.8604126
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 1 20 0.749367 0.751462 0.7940039
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 2 20 0.8223375 0.748538 0.7933591
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 7 3 20 0.8467121 0.7982456 0.8362347
## maxinter: 5 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 8 5 20 0.908838 0.744152 0.8546099
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 1 15 0.749367 0.751462 0.7940039
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 10 2 15 0.8178168 0.7821637 0.825274
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 11 3 15 0.8487646 0.7207602 0.8222115
## maxinter: 5 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 12 5 15 0.8952632 0.8654971 0.9000645
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 10 0.749367 0.751462 0.7940039
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 10 0.8153934 0.7843567 0.8170535
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 10 0.8542061 0.755848 0.8281754
## maxinter: 5 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 16 5 10 0.8892335 0.7573099 0.8647647
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 17 1 5 0.749367 0.751462 0.7940039
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 18 2 5 0.8138716 0.7865497 0.8065764
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 19 3 5 0.8581385 0.7836257 0.8420374
## maxinter: 5 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 20 5 5 0.8918295 0.7763158 0.8634752
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7493670 0.7514620 0.7940039
## 2 2 25 0.8187056 0.8347953 0.8568665
## 3 3 25 0.8596156 0.8011696 0.8596067
## 4 5 25 0.8878267 0.7456140 0.8604126
## 5 1 20 0.7493670 0.7514620 0.7940039
## 6 2 20 0.8223375 0.7485380 0.7933591
## 7 3 20 0.8467121 0.7982456 0.8362347
## 8 5 20 0.9088380 0.7441520 0.8546099
## 9 1 15 0.7493670 0.7514620 0.7940039
## 10 2 15 0.8178168 0.7821637 0.8252740
## 11 3 15 0.8487646 0.7207602 0.8222115
## 12 5 15 0.8952632 0.8654971 0.9000645
## 13 1 10 0.7493670 0.7514620 0.7940039
## 14 2 10 0.8153934 0.7843567 0.8170535
## 15 3 10 0.8542061 0.7558480 0.8281754
## 16 5 10 0.8892335 0.7573099 0.8647647
## 17 1 5 0.7493670 0.7514620 0.7940039
## 18 2 5 0.8138716 0.7865497 0.8065764
## 19 3 5 0.8581385 0.7836257 0.8420374
## 20 5 5 0.8918295 0.7763158 0.8634752
## maxinter nodesize rocTrain rocValid rocTest
## 12 5 15 0.8952632 0.8654971 0.9000645
##
## Observation 1 has a predicted value 0.32
## since this is the weighted average response across the 11 nodes it is a member of:
##
## 1) Node 137, containing 200 training observations, with node mean 0.16 and weight 0.248 :
## SER_countor <= 0.84
## Vr_post_1_countor <= 0.13
## V12 <= 44
##
## 2) Node 101, containing 54 training observations, with node mean 0.196 and weight 0.208 :
## SER_countor <= 0.71
## dce3SE9 <= 1.1
## 220 <= T2RGH_var
## lateSE12 <= 0.87
##
## 3) Node 34, containing 17 training observations, with node mean 0.588 and weight 0.132 :
## earlySE7 <= 1.5
## 0.21 <= texture_correlation_nondir_post1
## 26 <= V3
## 1.3 <= texture_diffentropy_nondir_post2
## lateSE12 <= 1.1
##
## 4) Node 68, containing 30 training observations, with node mean 0.742 and weight 0.104 :
## earlySE7 <= 1.5
## SER_countor <= 0.71
## 350 <= T2RGH_var
## A_inside <= 1.6
## 15 <= V3
##
## 5) Node 119, containing 78 training observations, with node mean 0.525 and weight 0.0607 :
## SER_countor <= 0.71
## 350 <= T2RGH_var
## 1.7 <= texture_sumentropy_nondir_post3
## 5.7 <= V12
## lateSE12 <= 1.2
##
## 6) Node 138, containing 206 training observations, with node mean 0.175 and weight 0.0605 :
## V8 <= 23
## texture_correlation_nondir_post1 <= 0.6
## SER_countor <= 0.76
## iAUC1_inside <= 1600
##
## 7) Node 105, containing 64 training observations, with node mean 0.188 and weight 0.0564 :
## 350 <= T2RGH_var
## 380 <= T2max_F_r_i
## texture_diffentropy_nondir_post2 <= 1.5
## texture_correlation_nondir_post1 <= 0.3
## 0.51 <= earlySE7
##
## 8) Node 135, containing 141 training observations, with node mean 0.148 and weight 0.0458 :
## Vr_post_1_countor <= 0.11
## 0.52 <= dce3SE13
## texture_diffentropy_nondir_post2 <= 1.6
## ave_T25 <= 230
## SER_countor <= 1
##
## 9) Node 48, containing 22 training observations, with node mean 0.364 and weight 0.039 :
## earlySE7 <= 1.4
## Vr_post_1_countor <= 0.14
## SER_countor <= 0.73
## dce3SE6 <= 0.71
## 8.6 <= V8
##
## 10) Node 139, containing 233 training observations, with node mean 0.186 and weight 0.0346 :
## texture_correlation_nondir_post1 <= 0.6
## earlySE7 <= 1.1
## SER_countor <= 0.84
## beta_inside <= 0.13
## V12 <= 30
##
## 11) Node 140, containing 572 training observations, with node mean 0.395 and weight 0.01 :
## ROOT NODE
## id C NC pred obs
## 1 33 0.3312089 0.6687911 NC C
## 2 34 0.4493398 0.5506602 NC C
## 3 52 0.3282720 0.6717280 NC NC
## 4 83 0.3296496 0.6703504 NC C
## 5 93 0.2786594 0.7213406 NC NC
## 6 96 0.3766474 0.6233526 NC NC
## id C NC pred obs
## 1 33 0.3712180 0.6287820 NC C
## 2 34 0.5841141 0.4158859 C C
## 3 52 0.3583589 0.6416411 NC NC
## 4 83 0.2580179 0.7419821 NC C
## 5 93 0.3866790 0.6133210 NC NC
## 6 96 0.4190312 0.5809688 NC NC
## id C NC pred obs
## 1 33 0.1791265 0.8208735 NC C
## 2 34 0.2115294 0.7884706 NC C
## 3 52 0.6009345 0.3990655 C NC
## 4 83 0.2706627 0.7293373 NC C
## 5 93 0.2148824 0.7851176 NC NC
## 6 96 0.2466795 0.7533205 NC NC
## id C NC pred obs
## 1 33 0.3201640 0.6798360 NC C
## 2 34 0.3831506 0.6168494 NC C
## 3 52 0.6283705 0.3716295 C NC
## 4 83 0.4205024 0.5794976 NC C
## 5 93 0.2223233 0.7776767 NC NC
## 6 96 0.3530334 0.6469666 NC NC
##
## Call:
## roc.default(response = perf_imgT2$obs, predictor = perf_imgT2$C)
##
## Data: perf_imgT2$C in 145 controls (perf_imgT2$obs C) > 212 cases (perf_imgT2$obs NC).
## Area under the curve: 0.7618
##
## Call:
## roc.default(response = perf_allT2$obs, predictor = perf_allT2$C)
##
## Data: perf_allT2$C in 135 controls (perf_allT2$obs C) > 222 cases (perf_allT2$obs NC).
## Area under the curve: 0.7757
##
## Call:
## roc.default(response = perf_imgT1$obs, predictor = perf_imgT1$C)
##
## Data: perf_imgT1$C in 156 controls (perf_imgT1$obs C) > 201 cases (perf_imgT1$obs NC).
## Area under the curve: 0.8186
##
## Call:
## roc.default(response = perf_all$obs, predictor = perf_all$C)
##
## Data: perf_all$C in 149 controls (perf_all$obs C) > 208 cases (perf_all$obs NC).
## Area under the curve: 0.8628
## Area under the curve: 0.7618
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.36445 0.7448 0.8069 0.869 0.5236 0.5896 0.6557
## Area under the curve: 0.7757
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.3702956 0.7926 0.8519 0.9111 0.518 0.5856 0.6532
## Area under the curve: 0.8186
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.4177992 0.6603 0.7308 0.8013 0.7612 0.8159 0.8706
## Area under the curve: 0.8628
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.3770788 0.745 0.8121 0.8725 0.7356 0.7885 0.8413
## massB massM nonmassB nonmassM
## 224 151 128 70
## massB massM nonmassB nonmassM
## 16 17 14 7
## massB massM nonmassB nonmassM
## 224 151 128 70
## massB massM nonmassB nonmassM
## 16 17 14 7
## massB massM nonmassB nonmassM
## 224 151 128 70
## massB massM nonmassB nonmassM
## 16 17 14 7
## -0.05882353 0.05
## Selected features for group: MeanDecreaseGini imgT2
## =========NULL
## [1] "T2RGH_mean" "ave_T210"
## [3] "T2texture_energy_nondir" "T2grad_margin_var"
## [5] "T2kurt_F_r_i" "T2texture_inversediffmoment_nondir"
## [7] "ave_T211" "ave_T213"
## [9] "T2texture_contrast_nondir" "ave_T214"
## [11] "ave_T28" "ave_T216"
## [13] "T2texture_sumentropy_nondir" "ave_T29"
## [15] "ave_T23" "T2_lesionSI"
## [17] "T2texture_correlation_nondir" "ave_T27"
## [19] "ave_T217" "T2grad_margin"
## [21] "T2max_F_r_i" "ave_T22"
## [23] "ave_T25" "T2skew_F_r_i"
## [25] "T2var_F_r_i" "T2_lesionSIstd"
## -0.065 0.05
## Selected features for group: MeanDecreaseGini allT2
## =========NULL
## [1] "T2RGH_var" "T2texture_entropy_nondir"
## [3] "T2texture_correlation_nondir" "T2RGH_mean"
## [5] "T2kurt_F_r_i" "T2wSI_predicted"
## [7] "ave_T210" "T2_lesionSIstd"
## [9] "T2max_F_r_i" "T2texture_sumaverage_nondir"
## [11] "ave_T27" "ave_T213"
## [13] "ave_T20" "ave_T217"
## [15] "T2texture_diffentropy_nondir" "T2grad_margin"
## [17] "ave_T24" "ave_T215"
## [19] "ave_T219" "T2mean_F_r_i"
## [21] "T2texture_inversediffmoment_nondir" "T2texture_sumentropy_nondir"
## [23] "ave_T29" "ave_T214"
## [25] "T2skew_F_r_i" "LMSIR_predicted"
## [27] "ave_T26" "T2min_F_r_i"
## [29] "ave_T25" "T2_lesionSI"
## 0.04347826 0.05
## Selected features for group: MeanDecreaseGini imgT1
## =========NULL
## [1] "texture_variance_nondir_post1" "SER_inside"
## [3] "texture_inversediffmoment_nondir_post4" "irregularity"
## [5] "V10" "texture_sumvariance_nondir_post2"
## [7] "max_RGH_var" "maxVr_countor"
## [9] "earlySE10" "beta_inside"
## [11] "Slope_ini_countor" "V2"
## [13] "earlySE12" "texture_energy_nondir_post3"
## [15] "dce2SE14" "texture_sumentropy_nondir_post3"
## [17] "texture_sumvariance_nondir_post3" "lateSE4"
## [19] "lateSE17" "beta_countor"
## [21] "earlySE17" "texture_contrast_nondir_post4"
## [23] "dce3SE1" "kurt_F_r_i"
## [25] "max_RGH_mean" "maxCr_inside"
## [27] "iiiMax_Margin_Gradient" "dce2SE2"
## [29] "earlySE9" "lateSE18"
## [31] "V8" "texture_energy_nondir_post1"
## [33] "peakCr_countor" "lateSE13"
## [35] "V13" "dce3SE0"
## [37] "A_inside"
## -0.05960265 0.05
## Selected features for group: MeanDecreaseGini all
## =========NULL
## [1] "irregularity" "alpha_inside"
## [3] "Vr_post_1_countor" "Tpeak_inside"
## [5] "SER_inside" "texture_sumaverage_nondir_post4"
## [7] "texture_sumvariance_nondir_post3" "edge_sharp_std"
## [9] "V10" "T2texture_variance_nondir"
## [11] "texture_sumentropy_nondir_post4" "V13"
## [13] "dce3SE4" "var_F_r_i"
## [15] "dce2SE18" "dce2SE19"
## [17] "dce2SE11" "dce3SE12"
## [19] "T2RGH_var" "ave_T212"
## [21] "ave_T22" "dce2SE12"
## [23] "dce2SE3" "dce3SE5"
## [25] "dce3SE7" "T2texture_energy_nondir"
## [27] "V14" "texture_contrast_nondir_post4"
## [29] "lateSE6" "A_inside"
## lesion_id cad_pt_no_txt exam_a_number_txt BIRADS lesion_label lesion_diagnosis
## 11 11 0111 6907205 4 nonmassB DUCT PAPILLOMA
## 12 12 0114 6896014 4 massM InvasiveDuctal
## 82 82 0409 5161803 4 massB BENIGN BREAST TISSUE
## 91 91 0463 7626269 4 massB FLORID HYPERPLASIA
## 110 110 0576 6905042 4 nonmassB BENIGN BREAST TISSUE
## 115 115 0603 4593568 4 nonmassB FIBROSIS
## 118 118 0613 4681594 4 nonmassM InsituDuctal
## 119 119 0613 4681594 3 nonmassB BENIGN BREAST TISSUE
## 135 135 0673 4585908 4 massB FIBROCYSTIC
## 138 138 0682 5050826 6 nonmassM InsituDuctal
## 145 145 0689 5205923 2 massB COLUMNAR CELL CHANGES
## 149 149 0691 5178056 5 massM InvasiveDuctal
## 150 150 0691 5178056 5 massM InvasiveDuctal
## 168 168 0721 4961869 6 massM InvasiveDuctal
## 178 178 0729 4805710 4 massB ATYPICAL LOBULAR HYPERPLASIA
## 182 182 0735 5276000 5 massM InvasiveDuctal
## 186 186 0742 5329785 4 massB DUCT PAPILLOMA
## 187 187 0742 5329785 4 nonmassB DUCT PAPILLOMA
## 189 189 0744 4848278 5 massM InvasiveDuctal
## 191 191 0748 4940559 6 massM ATYPICAL DUCTAL HYPERPLASIA
## 195 195 0755 5059877 4 nonmassM InsituDuctal
## 196 196 0755 5059877 4 nonmassB BENIGN BREAST TISSUE
## 252 252 0837 4559849 5 massM InvasiveDuctal
## 276 276 0861 5053396 5 massM InvasiveDuctal
## 283 283 0870 5141888 6 nonmassM InvasiveLobular
## 302 302 0887 6794529 4 massB FIBROCYSTIC
## 316 316 0943 5395204 4 nonmassM InsituDuctal
## 317 317 0943 5395204 4 nonmassB FIBROCYSTIC
## 320 320 0950 6931716 5 massM InvasiveDuctal
## 321 321 0950 6931716 5 nonmassM InvasiveDuctal
## 338 338 1004 6801264 4 nonmassB FIBROCYSTIC
## 363 363 1078 7105247 4 nonmassB DENSE FIBROSIS
## 364 364 1078 7105247 4 nonmassB DENSE FIBROSIS
## 377 377 2007 7366811 4 nonmassB FIBROADENOMA
## 395 395 2051 6712632 6 massM InsituDuctal
## 396 396 2051 6712632 6 nonmassB FIBROCYSTIC
## 397 397 2053 6776964 6 massM InvasiveDuctal
## 423 423 3021 7019819 4 massB BENIGN BREAST TISSUE
## 430 430 3033 5016967 5 massM InvasiveDuctal
## 456 456 3073 7043941 6 nonmassM IN SITU PAPILLARY CARCINOMA
## 499 499 4044 7066571 4 massB FIBROADENOMATOID
## 500 500 4044 7066571 4 massB FOCAL CELLULAR STROMA
## 501 501 4044 7066571 4 massB FIBROADENOMA
## 502 502 4045 7092118 4 massB FIBROADENOMA
## 516 516 6017 5086121 6 massM InvasiveLobular
## 517 517 6017 5086121 2 massB FIBROADENOMA
## 558 558 6044 5078981 5 massB FIBROCYSTIC
## 559 559 6044 5078981 5 massM InvasiveDuctal
## 583 583 6117 5154282 3 massB FIBROADENOMA
## 584 584 6141 7044114 2 massB FIBROADENOMA
## 598 598 7029 7014263 4 nonmassB ATYPICAL DUCTAL HYPERPLASIA
## 612 612 7086 6938067 4 massM InsituDuctal
## 613 613 7088 7066921 3 nonmassB FIBROCYSTIC
## 621 621 7105 7837892 4 massM InvasiveDuctal
## find_t2_signal_int
## 11 Hypointense or not seen
## 12 None
## 82 None
## 91 Hypointense or not seen
## 110 Hypointense or not seen
## 115 None
## 118 None
## 119 None
## 135 Hypointense or not seen
## 138 None
## 145 Hyperintense
## 149 Hypointense or not seen
## 150 Hypointense or not seen
## 168 Hyperintense
## 178 None
## 182 None
## 186 None
## 187 None
## 189 None
## 191 None
## 195 None
## 196 None
## 252 None
## 276 Hypointense or not seen
## 283 None
## 302 Hypointense or not seen
## 316 None
## 317 Hypointense or not seen
## 320 Hypointense or not seen
## 321 None
## 338 Hypointense or not seen
## 363 None
## 364 None
## 377 Hypointense or not seen
## 395 None
## 396 None
## 397 None
## 423 None
## 430 None
## 456 None
## 499 Hyperintense
## 500 Hyperintense
## 501 Slightly hyperintense
## 502 None
## 516 None
## 517 Hyperintense
## 558 None
## 559 None
## 583 Hyperintense
## 584 Hyperintense
## 598 None
## 612 None
## 613 None
## 621 Slightly hyperintense
##
## ============ bagging trees treedata_imgT2
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6631852 0.7243056 0.700491
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.7307178 0.7340278 0.7528642
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.7776571 0.7805556 0.7639935
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 4 1 20 0.6631852 0.7243056 0.700491
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 2 20 0.7182679 0.75 0.7292962
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 3 20 0.7928257 0.7055556 0.7368249
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 7 1 15 0.6631852 0.7243056 0.700491
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 8 2 15 0.7392855 0.7652778 0.7499182
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 3 15 0.7918423 0.7819444 0.7690671
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 10 1 10 0.6631852 0.7243056 0.700491
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 11 2 10 0.7349599 0.7284722 0.7545008
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 12 3 10 0.8031738 0.7590278 0.7559738
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 5 0.6631852 0.7243056 0.700491
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 5 0.7394462 0.7527778 0.7428805
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 5 0.7908011 0.7340278 0.7890344
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6631852 0.7243056 0.7004910
## 2 2 25 0.7307178 0.7340278 0.7528642
## 3 3 25 0.7776571 0.7805556 0.7639935
## 4 1 20 0.6631852 0.7243056 0.7004910
## 5 2 20 0.7182679 0.7500000 0.7292962
## 6 3 20 0.7928257 0.7055556 0.7368249
## 7 1 15 0.6631852 0.7243056 0.7004910
## 8 2 15 0.7392855 0.7652778 0.7499182
## 9 3 15 0.7918423 0.7819444 0.7690671
## 10 1 10 0.6631852 0.7243056 0.7004910
## 11 2 10 0.7349599 0.7284722 0.7545008
## 12 3 10 0.8031738 0.7590278 0.7559738
## 13 1 5 0.6631852 0.7243056 0.7004910
## 14 2 5 0.7394462 0.7527778 0.7428805
## 15 3 5 0.7908011 0.7340278 0.7890344
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 5 0.7908011 0.7340278 0.7890344
##
## Observation 1 has a predicted value 0.229
## since this is the weighted average response across the 10 nodes it is a member of:
##
## 1) Node 13, containing 18 training observations, with node mean 0.111 and weight 0.196 :
## 22 <= T2RGH_mean
## 0.00048 <= T2texture_energy_nondir
## T2texture_inversediffmoment_nondir <= 0.067
##
## 2) Node 7, containing 15 training observations, with node mean 0.0667 and weight 0.168 :
## 540 <= T2max_F_r_i
## 4800 <= T2grad_margin_var
## ave_T210 <= 100
##
## 3) Node 52, containing 256 training observations, with node mean 0.328 and weight 0.142 :
## T2texture_correlation_nondir <= 0.22
## 0.00039 <= T2texture_energy_nondir
##
## 4) Node 53, containing 262 training observations, with node mean 0.383 and weight 0.121 :
## 22 <= T2RGH_mean
## 41 <= ave_T210
## T2texture_correlation_nondir <= 0.26
##
## 5) Node 38, containing 93 training observations, with node mean 0.309 and weight 0.099 :
## 54 <= T2_lesionSIstd
## 22 <= T2RGH_mean
## 6400 <= T2grad_margin_var
##
## 6) Node 49, containing 180 training observations, with node mean 0.383 and weight 0.0964 :
## 0.00048 <= T2texture_energy_nondir
## 85 <= ave_T213
## 75 <= T2_lesionSIstd
##
## 7) Node 16, containing 20 training observations, with node mean 0.05 and weight 0.0744 :
## 22 <= T2RGH_mean
## ave_T210 <= 71
## T2kurt_F_r_i <= 1.4
##
## 8) Node 50, containing 199 training observations, with node mean 0.422 and weight 0.0487 :
## 68 <= T2_lesionSIstd
## T2kurt_F_r_i <= 1.2
##
## 9) Node 28, containing 51 training observations, with node mean 0.157 and weight 0.0439 :
## T2texture_correlation_nondir <= 0.26
## ave_T210 <= 72
##
## 10) Node 58, containing 569 training observations, with node mean 0.386 and weight 0.01 :
## ROOT NODE
##
## ============ bagging trees treedata_allT2
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6783088 0.65 0.5919987
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.7605538 0.7243056 0.6639132
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8018305 0.6923611 0.7066731
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 4 1 20 0.6783088 0.65 0.5919987
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 2 20 0.7584713 0.7402778 0.6733074
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 3 20 0.8002301 0.7277778 0.7277292
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 7 1 15 0.6783088 0.65 0.5919987
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 8 2 15 0.751594 0.7340278 0.6836735
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 3 15 0.8135284 0.6597222 0.6852932
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 10 1 10 0.6783088 0.65 0.5919987
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 11 2 10 0.7591333 0.7111111 0.6678005
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 12 3 10 0.810726 0.7243056 0.7095886
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 5 0.6783088 0.65 0.5919987
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 5 0.7535222 0.7069444 0.6451247
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 5 0.810289 0.7222222 0.7215743
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6783088 0.6500000 0.5919987
## 2 2 25 0.7605538 0.7243056 0.6639132
## 3 3 25 0.8018305 0.6923611 0.7066731
## 4 1 20 0.6783088 0.6500000 0.5919987
## 5 2 20 0.7584713 0.7402778 0.6733074
## 6 3 20 0.8002301 0.7277778 0.7277292
## 7 1 15 0.6783088 0.6500000 0.5919987
## 8 2 15 0.7515940 0.7340278 0.6836735
## 9 3 15 0.8135284 0.6597222 0.6852932
## 10 1 10 0.6783088 0.6500000 0.5919987
## 11 2 10 0.7591333 0.7111111 0.6678005
## 12 3 10 0.8107260 0.7243056 0.7095886
## 13 1 5 0.6783088 0.6500000 0.5919987
## 14 2 5 0.7535222 0.7069444 0.6451247
## 15 3 5 0.8102890 0.7222222 0.7215743
## maxinter nodesize rocTrain rocValid rocTest
## 6 3 20 0.8002301 0.7277778 0.7277292
##
## Observation 1 has a predicted value 0.213
## since this is the weighted average response across the 12 nodes it is a member of:
##
## 1) Node 58, containing 0.5 training observations, with node mean 0.148 and weight 0.333 :
## T2RGH_var <= 350
## T2wSI_predicted in {Hyperintense,Hypointense or not seen,Slightly hyperintense}
## 53 <= T2_lesionSI
##
## 2) Node 63, containing 0.5 training observations, with node mean 0.198 and weight 0.155 :
## T2texture_entropy_nondir <= 3.5
## 42 <= T2texture_sumaverage_nondir
## T2wSI_predicted in {Hyperintense,Hypointense or not seen,Slightly hyperintense}
##
## 3) Node 69, containing 0.5 training observations, with node mean 0.302 and weight 0.0914 :
## 22 <= T2RGH_mean
## T2wSI_predicted in {Hyperintense,Hypointense or not seen,Slightly hyperintense}
## 85 <= ave_T213
##
## 4) Node 71, containing 0.5 training observations, with node mean 0.312 and weight 0.0914 :
## T2wSI_predicted in {Hyperintense,Hypointense or not seen,Slightly hyperintense}
## T2texture_entropy_nondir <= 3.5
## 22 <= T2RGH_mean
##
## 5) Node 31, containing 0.5 training observations, with node mean 0.113 and weight 0.0774 :
## T2wSI_predicted in {Hyperintense,Hypointense or not seen,Slightly hyperintense}
## T2texture_correlation_nondir <= 0.11
##
## 6) Node 74, containing 295 training observations, with node mean 0.298 and weight 0.0718 :
## T2texture_entropy_nondir <= 3.4
## 78 <= T2_lesionSI
##
## 7) Node 29, containing 0.5 training observations, with node mean 0.208 and weight 0.0583 :
## T2wSI_predicted in {Hyperintense,Hypointense or not seen,Slightly hyperintense}
## 22 <= T2RGH_mean
## 240 <= ave_T217
##
## 8) Node 53, containing 113 training observations, with node mean 0.301 and weight 0.0441 :
## 460 <= T2max_F_r_i
## 56 <= T2grad_margin
## T2texture_entropy_nondir <= 3.5
##
## 9) Node 51, containing 104 training observations, with node mean 0.295 and weight 0.0356 :
## T2texture_entropy_nondir <= 3.5
## 62 <= T2texture_sumaverage_nondir
## 140 <= ave_T217
##
## 10) Node 59, containing 0.5 training observations, with node mean 0.159 and weight 0.0219 :
## T2wSI_predicted in {Hyperintense,Hypointense or not seen,Slightly hyperintense}
## T2RGH_var <= 350
## 53 <= T2_lesionSI
##
## 11) Node 75, containing 569 training observations, with node mean 0.386 and weight 0.01 :
## ROOT NODE
##
## 12) Node 46, containing 85 training observations, with node mean 0.322 and weight 0.00967 :
## 22 <= T2RGH_mean
## 1.5 <= T2texture_diffentropy_nondir
## T2texture_correlation_nondir <= 0.15
##
## ============ bagging trees treedata_imgT1
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.797402 0.8263889 0.8712585
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.8450933 0.7777778 0.8647959
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8779245 0.775 0.8741497
## maxinter: 5 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 4 5 25 0.9185456 0.7444444 0.8590136
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 1 20 0.797402 0.8263889 0.8712585
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 2 20 0.8391094 0.7631944 0.8685374
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 7 3 20 0.8703144 0.7930556 0.8814626
## maxinter: 5 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 8 5 20 0.9237132 0.7763889 0.880102
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 1 15 0.797402 0.8263889 0.8712585
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 10 2 15 0.8358057 0.7444444 0.8564626
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 11 3 15 0.8767804 0.7736111 0.8782313
## maxinter: 5 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 12 5 15 0.9093994 0.6972222 0.8370748
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 10 0.797402 0.8263889 0.8712585
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 10 0.8397779 0.7902778 0.877551
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 10 0.8715356 0.7555556 0.8620748
## maxinter: 5 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 16 5 10 0.9141686 0.7638889 0.8598639
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 17 1 5 0.798469 0.8263889 0.8717687
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 18 2 5 0.8297833 0.8104167 0.8828231
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 19 3 5 0.8741966 0.7555556 0.8755102
## maxinter: 5 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 20 5 5 0.9113919 0.7541667 0.8646259
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7974020 0.8263889 0.8712585
## 2 2 25 0.8450933 0.7777778 0.8647959
## 3 3 25 0.8779245 0.7750000 0.8741497
## 4 5 25 0.9185456 0.7444444 0.8590136
## 5 1 20 0.7974020 0.8263889 0.8712585
## 6 2 20 0.8391094 0.7631944 0.8685374
## 7 3 20 0.8703144 0.7930556 0.8814626
## 8 5 20 0.9237132 0.7763889 0.8801020
## 9 1 15 0.7974020 0.8263889 0.8712585
## 10 2 15 0.8358057 0.7444444 0.8564626
## 11 3 15 0.8767804 0.7736111 0.8782313
## 12 5 15 0.9093994 0.6972222 0.8370748
## 13 1 10 0.7974020 0.8263889 0.8712585
## 14 2 10 0.8397779 0.7902778 0.8775510
## 15 3 10 0.8715356 0.7555556 0.8620748
## 16 5 10 0.9141686 0.7638889 0.8598639
## 17 1 5 0.7984690 0.8263889 0.8717687
## 18 2 5 0.8297833 0.8104167 0.8828231
## 19 3 5 0.8741966 0.7555556 0.8755102
## 20 5 5 0.9113919 0.7541667 0.8646259
## maxinter nodesize rocTrain rocValid rocTest
## 18 2 5 0.8297833 0.8104167 0.8828231
##
## Observation 1 has a predicted value 0.214
## since this is the weighted average response across the 15 nodes it is a member of:
##
## 1) Node 37, containing 165 training observations, with node mean 0.133 and weight 0.179 :
## irregularity <= 0.94
## SER_inside <= 0.83
##
## 2) Node 47, containing 237 training observations, with node mean 0.169 and weight 0.128 :
## irregularity <= 0.98
## SER_inside <= 0.77
##
## 3) Node 51, containing 302 training observations, with node mean 0.344 and weight 0.111 :
## texture_variance_nondir_post1 <= 190
## 0.13 <= texture_inversediffmoment_nondir_post4
##
## 4) Node 48, containing 257 training observations, with node mean 0.167 and weight 0.0944 :
## SER_inside <= 0.8
## texture_variance_nondir_post1 <= 130
##
## 5) Node 49, containing 271 training observations, with node mean 0.351 and weight 0.0842 :
## texture_variance_nondir_post1 <= 170
## 0.68 <= SER_inside
##
## 6) Node 53, containing 375 training observations, with node mean 0.24 and weight 0.0834 :
## irregularity <= 0.98
## texture_variance_nondir_post1 <= 190
##
## 7) Node 46, containing 232 training observations, with node mean 0.159 and weight 0.079 :
## SER_inside <= 0.76
## texture_variance_nondir_post1 <= 140
##
## 8) Node 55, containing 395 training observations, with node mean 0.246 and weight 0.0596 :
## texture_variance_nondir_post1 <= 190
## irregularity <= 0.98
##
## 9) Node 52, containing 318 training observations, with node mean 0.217 and weight 0.0371 :
## irregularity <= 0.98
## SER_inside <= 0.82
##
## 10) Node 39, containing 183 training observations, with node mean 0.158 and weight 0.0333 :
## SER_inside <= 0.78
## 0.55 <= max_RGH_mean
##
## 11) Node 43, containing 213 training observations, with node mean 0.16 and weight 0.0324 :
## irregularity <= 0.98
## SER_inside <= 0.74
##
## 12) Node 54, containing 379 training observations, with node mean 0.24 and weight 0.0286 :
## irregularity <= 0.98
## texture_variance_nondir_post1 <= 190
##
## 13) Node 44, containing 217 training observations, with node mean 0.224 and weight 0.0211 :
## earlySE12 <= 0.81
## 0.42 <= earlySE10
##
## 14) Node 50, containing 274 training observations, with node mean 0.193 and weight 0.0186 :
## irregularity <= 0.98
## SER_inside <= 0.8
##
## 15) Node 56, containing 573 training observations, with node mean 0.386 and weight 0.01 :
## ROOT NODE
##
## ============ bagging trees treedata_all
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7976784 0.83125 0.8195161
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.8489305 0.7715278 0.8027419
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8747815 0.7638889 0.8130645
## maxinter: 5 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 4 5 25 0.9168873 0.7402778 0.8203226
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 1 20 0.7976784 0.83125 0.8195161
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 2 20 0.8502545 0.7569444 0.7935484
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 7 3 20 0.8644398 0.7388889 0.7967742
## maxinter: 5 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 8 5 20 0.9072784 0.7430556 0.8085484
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 1 15 0.7979034 0.8381944 0.8262903
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 10 2 15 0.8417254 0.7618056 0.7975806
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 11 3 15 0.8733032 0.7430556 0.7946774
## maxinter: 5 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 12 5 15 0.9089881 0.7666667 0.8275806
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 10 0.7976784 0.83125 0.8195161
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 10 0.8420339 0.7875 0.8224194
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 10 0.8730268 0.7083333 0.7690323
## maxinter: 5 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 16 5 10 0.9096115 0.7819444 0.83
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 17 1 5 0.7976784 0.8381944 0.8246774
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 18 2 5 0.8392894 0.7631944 0.8003226
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 19 3 5 0.8800391 0.7743056 0.8079032
## maxinter: 5 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 20 5 5 0.9032227 0.7361111 0.8154839
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7976784 0.8312500 0.8195161
## 2 2 25 0.8489305 0.7715278 0.8027419
## 3 3 25 0.8747815 0.7638889 0.8130645
## 4 5 25 0.9168873 0.7402778 0.8203226
## 5 1 20 0.7976784 0.8312500 0.8195161
## 6 2 20 0.8502545 0.7569444 0.7935484
## 7 3 20 0.8644398 0.7388889 0.7967742
## 8 5 20 0.9072784 0.7430556 0.8085484
## 9 1 15 0.7979034 0.8381944 0.8262903
## 10 2 15 0.8417254 0.7618056 0.7975806
## 11 3 15 0.8733032 0.7430556 0.7946774
## 12 5 15 0.9089881 0.7666667 0.8275806
## 13 1 10 0.7976784 0.8312500 0.8195161
## 14 2 10 0.8420339 0.7875000 0.8224194
## 15 3 10 0.8730268 0.7083333 0.7690323
## 16 5 10 0.9096115 0.7819444 0.8300000
## 17 1 5 0.7976784 0.8381944 0.8246774
## 18 2 5 0.8392894 0.7631944 0.8003226
## 19 3 5 0.8800391 0.7743056 0.8079032
## 20 5 5 0.9032227 0.7361111 0.8154839
## maxinter nodesize rocTrain rocValid rocTest
## 16 5 10 0.9096115 0.7819444 0.83
##
## Observation 1 has a predicted value 0.12
## since this is the weighted average response across the 12 nodes it is a member of:
##
## 1) Node 104, containing 51 training observations, with node mean 0.098 and weight 0.186 :
## texture_sumvariance_nondir_post3 <= 620
## Vr_post_1_countor <= 0.13
## A_inside <= 1.7
## alpha_inside <= 0.57
## T2RGH_var <= 580
##
## 2) Node 137, containing 110 training observations, with node mean 0.0818 and weight 0.184 :
## texture_sumvariance_nondir_post3 <= 620
## irregularity <= 0.93
## SER_inside <= 0.78
##
## 3) Node 138, containing 142 training observations, with node mean 0.0775 and weight 0.155 :
## SER_inside <= 0.76
## 0.00046 <= T2texture_energy_nondir
## Vr_post_1_countor <= 0.13
## var_F_r_i <= 83000
##
## 4) Node 145, containing 161 training observations, with node mean 0.106 and weight 0.106 :
## irregularity <= 0.98
## SER_inside <= 0.78
## var_F_r_i <= 26000
##
## 5) Node 147, containing 207 training observations, with node mean 0.217 and weight 0.0946 :
## irregularity <= 0.98
## texture_sumaverage_nondir_post4 <= 270
## texture_sumvariance_nondir_post3 <= 380
##
## 6) Node 146, containing 164 training observations, with node mean 0.165 and weight 0.0697 :
## edge_sharp_std <= 1
## T2RGH_var <= 330
## dce2SE3 <= 1.6
##
## 7) Node 140, containing 151 training observations, with node mean 0.151 and weight 0.061 :
## dce2SE12 <= 1
## irregularity <= 0.98
## lateSE6 <= 1.4
## alpha_inside <= 0.54
## T2texture_variance_nondir <= 820
##
## 8) Node 148, containing 224 training observations, with node mean 0.152 and weight 0.0584 :
## SER_inside <= 0.74
## A_inside <= 310
## var_F_r_i <= 84000
##
## 9) Node 143, containing 156 training observations, with node mean 0.109 and weight 0.0381 :
## SER_inside <= 0.72
## Vr_post_1_countor <= 0.14
##
## 10) Node 98, containing 47 training observations, with node mean 0.128 and weight 0.0189 :
## 5.9 <= Tpeak_inside
## Vr_post_1_countor <= 0.13
## A_inside <= 1.5
## irregularity <= 0.98
## dce3SE5 <= 1.1
##
## 11) Node 123, containing 70 training observations, with node mean 0.186 and weight 0.0188 :
## texture_sumvariance_nondir_post3 <= 830
## 0.68 <= SER_inside
## var_F_r_i <= 67000
## irregularity <= 0.86
##
## 12) Node 149, containing 573 training observations, with node mean 0.386 and weight 0.01 :
## ROOT NODE
## id C NC pred obs
## 1 11 0.2286915 0.7713085 NC NC
## 2 12 0.3732969 0.6267031 NC C
## 3 82 0.3902843 0.6097157 NC NC
## 4 91 0.3066730 0.6933270 NC NC
## 5 110 0.4656427 0.5343573 NC NC
## 6 115 0.4181730 0.5818270 NC NC
## id C NC pred obs
## 1 11 0.2127718 0.7872282 NC NC
## 2 12 0.4037416 0.5962584 NC C
## 3 82 0.4548350 0.5451650 NC NC
## 4 91 0.2873551 0.7126449 NC NC
## 5 110 0.4700913 0.5299087 NC NC
## 6 115 0.4833537 0.5166463 NC NC
## id C NC pred obs
## 1 11 0.2138674 0.7861326 NC NC
## 2 12 0.5889689 0.4110311 C C
## 3 82 0.5141820 0.4858180 C NC
## 4 91 0.2617287 0.7382713 NC NC
## 5 110 0.1790378 0.8209622 NC NC
## 6 115 0.2375824 0.7624176 NC NC
## id C NC pred obs
## 1 11 0.1204813 0.8795187 NC NC
## 2 12 0.5365965 0.4634035 C C
## 3 82 0.4484758 0.5515242 NC NC
## 4 91 0.3016373 0.6983627 NC NC
## 5 110 0.1778854 0.8221146 NC NC
## 6 115 0.3851544 0.6148456 NC NC
##
## Call:
## roc.default(response = perf_imgT2$obs, predictor = perf_imgT2$C)
##
## Data: perf_imgT2$C in 192 controls (perf_imgT2$obs C) > 277 cases (perf_imgT2$obs NC).
## Area under the curve: 0.7657
##
## Call:
## roc.default(response = perf_allT2$obs, predictor = perf_allT2$C)
##
## Data: perf_allT2$C in 184 controls (perf_allT2$obs C) > 285 cases (perf_allT2$obs NC).
## Area under the curve: 0.7584
##
## Call:
## roc.default(response = perf_imgT1$obs, predictor = perf_imgT1$C)
##
## Data: perf_imgT1$C in 198 controls (perf_imgT1$obs C) > 271 cases (perf_imgT1$obs NC).
## Area under the curve: 0.8301
##
## Call:
## roc.default(response = perf_all$obs, predictor = perf_all$C)
##
## Data: perf_all$C in 199 controls (perf_all$obs C) > 270 cases (perf_all$obs NC).
## Area under the curve: 0.8532
## Area under the curve: 0.7657
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.36445 0.7604 0.8177 0.8698 0.5415 0.5993 0.6606
## Area under the curve: 0.7584
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.432035 0.5489 0.6196 0.6902 0.7333 0.7825 0.8316
## Area under the curve: 0.8301
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.4174294 0.6566 0.7172 0.7778 0.786 0.8303 0.8745
## Area under the curve: 0.8532
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.4104014 0.6935 0.7538 0.8092 0.7852 0.8296 0.8704
## massB massM nonmassB nonmassM
## 223 149 127 69
## massB massM nonmassB nonmassM
## 17 19 15 8
## massB massM nonmassB nonmassM
## 223 149 127 69
## massB massM nonmassB nonmassM
## 17 19 15 8
## massB massM nonmassB nonmassM
## 223 149 127 69
## massB massM nonmassB nonmassM
## 17 19 15 8
## -0.06965174 0.05
## Selected features for group: MeanDecreaseGini imgT2
## =========NULL
## [1] "T2RGH_var" "T2texture_entropy_nondir"
## [3] "T2kurt_F_r_i" "T2grad_margin"
## [5] "ave_T27" "T2max_F_r_i"
## [7] "ave_T211" "ave_T213"
## [9] "ave_T212" "ave_T25"
## [11] "ave_T214" "T2mean_F_r_i"
## [13] "ave_T29" "T2texture_contrast_nondir"
## [15] "ave_T21" "ave_T215"
## [17] "ave_T22" "T2texture_sumvariance_nondir"
## [19] "ave_T217" "T2texture_energy_nondir"
## [21] "T2skew_F_r_i" "T2texture_correlation_nondir"
## [23] "T2_lesionSIstd" "T2texture_inversediffmoment_nondir"
## [25] "ave_T218" "ave_T219"
## [27] "T2texture_diffentropy_nondir" "ave_T28"
## [29] "ave_T24" "T2texture_sumaverage_nondir"
## [31] "T2texture_variance_nondir" "ave_T20"
## [33] "ave_T26" "ave_T210"
## [35] "T2_lesionSI"
## 0.05529954 0.05
## 0 0.05
## Selected features for group: MeanDecreaseGini allT2
## =========NULL
## [1] "T2texture_entropy_nondir" "T2RGH_var"
## [3] "T2kurt_F_r_i" "T2texture_correlation_nondir"
## [5] "ave_T210" "T2grad_margin_var"
## [7] "T2texture_energy_nondir" "T2max_F_r_i"
## [9] "LMSIR_predicted" "T2RGH_mean"
## [11] "ave_T213" "ave_T27"
## [13] "T2_lesionSIstd" "T2wSI_predicted"
## [15] "ave_T218" "T2texture_variance_nondir"
## [17] "ave_T23" "T2texture_inversediffmoment_nondir"
## [19] "ave_T28" "ave_T219"
## [21] "ave_T217" "T2mean_F_r_i"
## [23] "ave_T25" "ave_T214"
## [25] "ave_T216" "ave_T22"
## [27] "ave_T215" "T2texture_diffvariance_nondir"
## [29] "ave_T20" "T2texture_sumaverage_nondir"
## [31] "ave_T211" "T2min_F_r_i"
## [33] "ave_T26" "T2skew_F_r_i"
## [35] "T2_lesionSI"
## 0.08441558 0.05
## -0.0141844 0.05
## Selected features for group: MeanDecreaseGini imgT1
## =========NULL
## [1] "SER_inside" "texture_variance_nondir_post1"
## [3] "texture_diffvariance_nondir_post1" "texture_correlation_nondir_post2"
## [5] "V15" "V14"
## [7] "texture_diffentropy_nondir_post3" "Vr_increasingRate_countor"
## [9] "kurt_F_r_i" "V16"
## [11] "texture_sumaverage_nondir_post3" "texture_sumentropy_nondir_post4"
## [13] "iAUC1_countor" "earlySE15"
## [15] "lateSE4" "edge_sharp_mean"
## [17] "earlySE9" "lateSE14"
## [19] "V17" "dce3SE2"
## [21] "dce2SE16" "dce2SE1"
## [23] "V7" "dce3SE1"
## [25] "V3" "earlySE3"
## [27] "lateSE8" "Kpeak_inside"
## [29] "texture_contrast_nondir_post4" "texture_sumentropy_nondir_post1"
## [31] "lateSE6" "peakCr_countor"
## [33] "V2" "A_inside"
## 0.01234568 0.05
## Selected features for group: MeanDecreaseGini all
## =========NULL
## [1] "texture_sumvariance_nondir_post1" "irregularity"
## [3] "washoutRate_inside" "texture_energy_nondir_post4"
## [5] "texture_correlation_nondir_post4" "earlySE6"
## [7] "max_RGH_mean" "V2"
## [9] "skew_F_r_i" "iMax_Variance_uptake"
## [11] "maxVr_inside" "dce3SE0"
## [13] "texture_entropy_nondir_post2" "beta_countor"
## [15] "lateSE14" "T2grad_margin"
## [17] "T2texture_sumentropy_nondir" "texture_sumvariance_nondir_post4"
## [19] "ave_T213" "ave_T28"
## [21] "ave_T21" "dce3SE7"
## [23] "earlySE10" "ave_T26"
## [25] "V10" "T2mean_F_r_i"
## [27] "Kpeak_countor" "earlySE9"
## [29] "ivVariance" "ave_T210"
## [31] "earlySE3" "dce3SE17"
## [33] "mean_F_r_i" "V6"
## [35] "A_inside"
## lesion_id cad_pt_no_txt exam_a_number_txt BIRADS lesion_label
## 10 10 0102 4755778 4 massB
## 69 69 0277 5077098 5 nonmassM
## 72 72 0293 7491268 4 nonmassB
## 81 81 0388 7395410 5 massM
## 92 92 0465 4885863 2 nonmassB
## 95 95 0503 6697826 3 massM
## 101 101 0551 4804820 4 massB
## 108 108 0572 4681582 4 nonmassM
## 109 109 0573 5142109 4 nonmassB
## 128 128 0664 7081071 4 nonmassB
## 155 155 0707 5184832 4 nonmassB
## 156 156 0707 5184832 4 nonmassB
## 184 184 0737 4559808 3 massM
## 199 199 0764 5088503 5 massM
## 200 200 0764 5088503 5 massM
## 202 202 0767 5306672 4 massB
## 205 205 0775 5437780 3 massB
## 206 206 0775 5437780 3 nonmassB
## 207 207 0775 5437780 3 nonmassB
## 210 210 0782 4775699 5 massM
## 243 243 0827 4985128 4 massM
## 244 244 0827 4985128 4 nonmassM
## 245 245 0827 4985128 4 massB
## 247 247 0829 5264139 5 massM
## 255 255 0843 4798594 4 massB
## 256 256 0843 6792402 4 massB
## 274 274 0857 4870283 4 massB
## 275 275 0857 5013393 2 massM
## 289 289 0875 7141879 4 massB
## 290 290 0875 5396107 4 nonmassB
## 345 345 1021 6760795 4 massB
## 352 352 1050 7296806 3 nonmassB
## 370 370 1090 4288694 4 nonmassB
## 371 371 1092 4951061 6 massB
## 372 372 1092 4951061 4 nonmassB
## 381 381 2024 5190122 6 nonmassM
## 382 382 2027 5465838 6 massM
## 401 401 2068 7559583 5 massM
## 420 420 3017 7014437 4 nonmassB
## 425 425 3025 7103914 4 massB
## 428 428 3030 7642998 4 massB
## 454 454 3072 7054863 6 nonmassM
## 473 473 3093 7438787 4 massB
## 481 481 4017 6979356 4 massB
## 482 482 4017 6979356 4 massB
## 505 505 4055 7439091 4 massB
## 507 507 6004 ACC108249 6 massM
## 508 508 6004 ACC108249 5 nonmassM
## 544 544 6037 5043444 5 massM
## 545 545 6037 5043444 5 massM
## 549 549 6039 ACC109197 5 massM
## 550 550 6039 ACC109197 5 nonmassM
## 551 551 6039 ACC109197 5 massM
## 555 555 6042 4504274 3 massM
## 556 556 6042 4504274 3 massM
## 566 566 6047 5275305 6 nonmassM
## 582 582 6114 5148523 6 nonmassB
## 620 620 7104 6941351 5 massM
## 636 636 7220 7288789 4 nonmassB
## lesion_diagnosis find_t2_signal_int
## 10 FIBROCYSTIC Hyperintense
## 69 InsituDuctal Hypointense or not seen
## 72 BENIGN BREAST TISSUE None
## 81 InvasiveDuctal None
## 92 ATYPICAL DUCTAL HYPERPLASIA None
## 95 InvasiveDuctal Hyperintense
## 101 FIBROEPITHELIAL Slightly hyperintense
## 108 InvasiveDuctal None
## 109 COLUMNAR CELL CHANGES None
## 128 ATYPICAL DUCTAL HYPERPLASIA Hypointense or not seen
## 155 COMPLEX PAPILLARY LESION Hypointense or not seen
## 156 COMPLEX PAPILLARY LESION Hypointense or not seen
## 184 InsituDuctal Hyperintense
## 199 InvasiveDuctal None
## 200 InvasiveDuctal None
## 202 ATYPICAL PAPILLARY LESION None
## 205 BENIGN BREAST TISSUE Slightly hyperintense
## 206 BENIGN BREAST TISSUE Hypointense or not seen
## 207 BENIGN BREAST TISSUE Slightly hyperintense
## 210 InvasiveDuctal None
## 243 INSITUPAPILLARYCARCINOMA None
## 244 INSITUPAPILLARYCARCINOMA None
## 245 USUAL DUCTAL HYPERPLASIA None
## 247 InvasiveDuctal Hypointense or not seen
## 255 FIBROCYSTIC None
## 256 FIBROCYSTIC None
## 274 FIBROCYSTIC Hypointense or not seen
## 275 InsituDuctal Hypointense or not seen
## 289 DUCT PAPILLOMA None
## 290 PAPILLARY LESION None
## 345 FIBROADENOMA Hypointense or not seen
## 352 DENSE FIBROSIS None
## 370 ATYPICAL DUCTAL HYPERPLASIA Slightly hyperintense
## 371 InsituLobular None
## 372 InsituLobular None
## 381 InsituDuctal Hypointense or not seen
## 382 InsituDuctal Hyperintense
## 401 InvasiveDuctal Hypointense or not seen
## 420 FOCAL HYPERPLASIA Slightly hyperintense
## 425 FIBROADENOMA Hyperintense
## 428 BENIGN BREAST TISSUE Slightly hyperintense
## 454 InsituDuctal None
## 473 BENIGN BREAST TISSUE None
## 481 USUAL DUCTAL HYPERPLASIA Hypointense or not seen
## 482 USUAL DUCTAL HYPERPLASIA Hypointense or not seen
## 505 PSEUDOANGIOMATOUS STROMAL HYPERPLASIA None
## 507 InvasiveLobular Hypointense or not seen
## 508 InvasiveLobular None
## 544 InvasiveDuctal None
## 545 InvasiveDuctal None
## 549 InsituDuctal Hyperintense
## 550 InsituDuctal None
## 551 InsituDuctal Hyperintense
## 555 InvasiveDuctal Hypointense or not seen
## 556 InvasiveDuctal Hypointense or not seen
## 566 Adenocarcinoma None
## 582 BENIGN BREAST TISSUE None
## 620 InvasiveDuctal None
## 636 FIBROCYSTIC Hypointense or not seen
##
## ============ bagging trees treedata_imgT2
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6736632 0.6331019 0.6231366
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.7471363 0.625 0.6656832
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.7909699 0.634838 0.7049689
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 4 1 20 0.6736632 0.6331019 0.6231366
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 2 20 0.7479817 0.6614583 0.6993789
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 3 20 0.7969332 0.5677083 0.6849379
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 7 1 15 0.6736632 0.6331019 0.6231366
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 8 2 15 0.7488598 0.6047454 0.6641304
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 3 15 0.7913958 0.59375 0.6748447
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 10 1 10 0.6736632 0.6331019 0.6231366
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 11 2 10 0.7405898 0.6822917 0.7045031
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 12 3 10 0.790308 0.599537 0.6493789
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 5 0.6736632 0.6331019 0.6231366
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 5 0.735865 0.6221065 0.665528
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 5 0.7868021 0.5758102 0.6285714
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6736632 0.6331019 0.6231366
## 2 2 25 0.7471363 0.6250000 0.6656832
## 3 3 25 0.7909699 0.6348380 0.7049689
## 4 1 20 0.6736632 0.6331019 0.6231366
## 5 2 20 0.7479817 0.6614583 0.6993789
## 6 3 20 0.7969332 0.5677083 0.6849379
## 7 1 15 0.6736632 0.6331019 0.6231366
## 8 2 15 0.7488598 0.6047454 0.6641304
## 9 3 15 0.7913958 0.5937500 0.6748447
## 10 1 10 0.6736632 0.6331019 0.6231366
## 11 2 10 0.7405898 0.6822917 0.7045031
## 12 3 10 0.7903080 0.5995370 0.6493789
## 13 1 5 0.6736632 0.6331019 0.6231366
## 14 2 5 0.7358650 0.6221065 0.6655280
## 15 3 5 0.7868021 0.5758102 0.6285714
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.7909699 0.634838 0.7049689
##
## Observation 1 has a predicted value 0.275
## since this is the weighted average response across the 9 nodes it is a member of:
##
## 1) Node 31, containing 71 training observations, with node mean 0.169 and weight 0.245 :
## 61 <= T2_lesionSIstd
## T2texture_entropy_nondir <= 3.4
## 450 <= T2texture_contrast_nondir
##
## 2) Node 33, containing 75 training observations, with node mean 0.293 and weight 0.203 :
## 410 <= T2max_F_r_i
## T2RGH_var <= 350
## T2texture_sumvariance_nondir <= 1000
##
## 3) Node 46, containing 217 training observations, with node mean 0.332 and weight 0.184 :
## T2texture_entropy_nondir <= 3.4
## 58 <= T2mean_F_r_i
## 60 <= T2_lesionSIstd
##
## 4) Node 43, containing 183 training observations, with node mean 0.344 and weight 0.117 :
## 0.00054 <= T2texture_energy_nondir
## T2skew_F_r_i <= 1.4
## 55 <= T2_lesionSIstd
##
## 5) Node 36, containing 94 training observations, with node mean 0.191 and weight 0.105 :
## T2texture_entropy_nondir <= 3.4
## 380 <= T2texture_contrast_nondir
##
## 6) Node 25, containing 51 training observations, with node mean 0.314 and weight 0.0877 :
## 0.00054 <= T2texture_energy_nondir
## 67 <= T2texture_sumaverage_nondir
## 220 <= ave_T219
##
## 7) Node 48, containing 288 training observations, with node mean 0.383 and weight 0.031 :
## 0.00054 <= T2texture_energy_nondir
## ave_T25 <= 240
## 200 <= T2RGH_var
##
## 8) Node 40, containing 147 training observations, with node mean 0.537 and weight 0.0165 :
## 0.26 <= T2texture_correlation_nondir
## ave_T20 <= 200
##
## 9) Node 50, containing 564 training observations, with node mean 0.384 and weight 0.01 :
## ROOT NODE
##
## ============ bagging trees treedata_allT2
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6863303 0.5295139 0.6283701
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.7539777 0.5792824 0.6888787
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8048952 0.5665509 0.6758578
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 4 1 20 0.6863303 0.5295139 0.6283701
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 2 20 0.7557667 0.5891204 0.6979167
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 3 20 0.8098296 0.6064815 0.743413
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 7 1 15 0.6863303 0.5295139 0.6283701
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 8 2 15 0.7491743 0.5873843 0.6878064
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 3 15 0.7899672 0.5972222 0.7225797
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 10 1 10 0.6863303 0.5295139 0.6283701
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 11 2 10 0.7529358 0.5815972 0.6934743
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 12 3 10 0.8054063 0.5358796 0.6609988
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 5 0.6863303 0.5295139 0.6283701
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 5 0.7619069 0.5769676 0.6842831
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 5 0.8074509 0.4375 0.6761642
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6863303 0.5295139 0.6283701
## 2 2 25 0.7539777 0.5792824 0.6888787
## 3 3 25 0.8048952 0.5665509 0.6758578
## 4 1 20 0.6863303 0.5295139 0.6283701
## 5 2 20 0.7557667 0.5891204 0.6979167
## 6 3 20 0.8098296 0.6064815 0.7434130
## 7 1 15 0.6863303 0.5295139 0.6283701
## 8 2 15 0.7491743 0.5873843 0.6878064
## 9 3 15 0.7899672 0.5972222 0.7225797
## 10 1 10 0.6863303 0.5295139 0.6283701
## 11 2 10 0.7529358 0.5815972 0.6934743
## 12 3 10 0.8054063 0.5358796 0.6609988
## 13 1 5 0.6863303 0.5295139 0.6283701
## 14 2 5 0.7619069 0.5769676 0.6842831
## 15 3 5 0.8074509 0.4375000 0.6761642
## maxinter nodesize rocTrain rocValid rocTest
## 6 3 20 0.8098296 0.6064815 0.743413
##
## Observation 1 has a predicted value 0.232
## since this is the weighted average response across the 6 nodes it is a member of:
##
## 1) Node 52, containing 0.5 training observations, with node mean 0.381 and weight 0.324 :
## T2wSI_predicted in {Hyperintense,Hypointense or not seen,Slightly hyperintense}
## 22 <= T2RGH_mean
## 0.1 <= T2texture_correlation_nondir
##
## 2) Node 33, containing 63 training observations, with node mean 0.286 and weight 0.214 :
## 410 <= T2max_F_r_i
## T2grad_margin_var <= 4800
## T2RGH_var <= 350
##
## 3) Node 42, containing 0.5 training observations, with node mean 0.121 and weight 0.192 :
## T2wSI_predicted in {Hyperintense,Hypointense or not seen,Slightly hyperintense}
## T2RGH_var <= 350
## 51 <= T2mean_F_r_i
##
## 4) Node 41, containing 0.5 training observations, with node mean 0.151 and weight 0.122 :
## T2wSI_predicted in {Hyperintense,Hypointense or not seen}
## T2max_F_r_i <= 690
## -0.19 <= T2kurt_F_r_i
##
## 5) Node 58, containing 564 training observations, with node mean 0.384 and weight 0.01 :
## ROOT NODE
##
## 6) Node 50, containing 164 training observations, with node mean 0.457 and weight 0.00435 :
## 0.29 <= T2texture_correlation_nondir
## ave_T20 <= 300
##
## ============ bagging trees treedata_imgT1
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7733748 0.724537 0.7739394
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.8300524 0.666088 0.7593939
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8634993 0.7569444 0.8221212
## maxinter: 5 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 4 5 25 0.9056029 0.7152778 0.815
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 1 20 0.7733748 0.724537 0.7739394
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 2 20 0.830865 0.6597222 0.7507576
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 7 3 20 0.8499869 0.697338 0.7845455
## maxinter: 5 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 8 5 20 0.8927326 0.7430556 0.8169697
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 1 15 0.77327 0.7256944 0.7742424
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 10 2 15 0.8307471 0.7164352 0.7772727
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 11 3 15 0.8593119 0.7025463 0.7904545
## maxinter: 5 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 12 5 15 0.8996855 0.71875 0.8521212
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 10 0.7733748 0.724537 0.7739394
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 10 0.8280275 0.6753472 0.7675758
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 10 0.8672018 0.75 0.8221212
## maxinter: 5 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 16 5 10 0.899502 0.693287 0.7936364
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 17 1 5 0.7729751 0.7251157 0.7745455
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 18 2 5 0.8255505 0.6707176 0.7536364
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 19 3 5 0.8573526 0.681713 0.7734848
## maxinter: 5 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 20 5 5 0.8943185 0.7210648 0.8151515
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7733748 0.7245370 0.7739394
## 2 2 25 0.8300524 0.6660880 0.7593939
## 3 3 25 0.8634993 0.7569444 0.8221212
## 4 5 25 0.9056029 0.7152778 0.8150000
## 5 1 20 0.7733748 0.7245370 0.7739394
## 6 2 20 0.8308650 0.6597222 0.7507576
## 7 3 20 0.8499869 0.6973380 0.7845455
## 8 5 20 0.8927326 0.7430556 0.8169697
## 9 1 15 0.7732700 0.7256944 0.7742424
## 10 2 15 0.8307471 0.7164352 0.7772727
## 11 3 15 0.8593119 0.7025463 0.7904545
## 12 5 15 0.8996855 0.7187500 0.8521212
## 13 1 10 0.7733748 0.7245370 0.7739394
## 14 2 10 0.8280275 0.6753472 0.7675758
## 15 3 10 0.8672018 0.7500000 0.8221212
## 16 5 10 0.8995020 0.6932870 0.7936364
## 17 1 5 0.7729751 0.7251157 0.7745455
## 18 2 5 0.8255505 0.6707176 0.7536364
## 19 3 5 0.8573526 0.6817130 0.7734848
## 20 5 5 0.8943185 0.7210648 0.8151515
## maxinter nodesize rocTrain rocValid rocTest
## 12 5 15 0.8996855 0.71875 0.8521212
##
## Observation 1 has a predicted value 0.57
## since this is the weighted average response across the 9 nodes it is a member of:
##
## 1) Node 76, containing 61 training observations, with node mean 0.934 and weight 0.307 :
## 170 <= texture_variance_nondir_post1
## 0.92 <= lateSE6
## edge_sharp_mean <= 0.66
## texture_diffentropy_nondir_post3 <= 1.6
##
## 2) Node 63, containing 46 training observations, with node mean 0.196 and weight 0.239 :
## 0.81 <= SER_inside
## texture_correlation_nondir_post2 <= 0.72
## V3 <= 21
## V7 <= 7.4
## 180 <= texture_sumaverage_nondir_post3
##
## 3) Node 75, containing 61 training observations, with node mean 0.328 and weight 0.107 :
## 0.24 <= texture_correlation_nondir_post2
## 0.64 <= earlySE15
## V16 <= 8
##
## 4) Node 65, containing 48 training observations, with node mean 0.792 and weight 0.1 :
## 0.24 <= texture_correlation_nondir_post2
## 3.5 <= peakCr_countor
## 190 <= texture_variance_nondir_post1
##
## 5) Node 47, containing 32 training observations, with node mean 0.438 and weight 0.086 :
## 170 <= texture_variance_nondir_post1
## lateSE6 <= 1.1
##
## 6) Node 64, containing 47 training observations, with node mean 0.17 and weight 0.0809 :
## 0.095 <= Vr_increasingRate_countor
## texture_diffvariance_nondir_post1 <= 130
## V16 <= 12
## 1.3 <= texture_diffentropy_nondir_post3
##
## 7) Node 8, containing 10 training observations, with node mean 1 and weight 0.0479 :
## 190 <= texture_variance_nondir_post1
## V16 <= 14
## texture_sumentropy_nondir_post1 <= 1.8
##
## 8) Node 90, containing 98 training observations, with node mean 0.857 and weight 0.021 :
## 0.68 <= SER_inside
## 170 <= texture_variance_nondir_post1
## texture_diffvariance_nondir_post1 <= 310
## 0.59 <= earlySE15
##
## 9) Node 100, containing 568 training observations, with node mean 0.384 and weight 0.01 :
## ROOT NODE
##
## ============ bagging trees treedata_all
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.8094037 0.3570602 0.7030303
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.8403277 0.6724537 0.7498485
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8680931 0.6915509 0.7816667
## maxinter: 5 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 4 5 25 0.9043775 0.6568287 0.7604545
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 1 20 0.8094037 0.3570602 0.7030303
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 2 20 0.8441743 0.6603009 0.7313636
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 7 3 20 0.878401 0.6452546 0.7421212
## maxinter: 5 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 8 5 20 0.8981848 0.65625 0.7927273
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 1 15 0.8094037 0.3570602 0.7030303
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 10 2 15 0.8380603 0.6278935 0.7131818
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 11 3 15 0.8678637 0.6811343 0.7371212
## maxinter: 5 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 12 5 15 0.9175098 0.666088 0.8127273
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 10 0.8094037 0.3570602 0.7030303
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 10 0.8471101 0.6730324 0.7534848
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 10 0.8846003 0.666088 0.7825758
## maxinter: 5 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 16 5 10 0.916363 0.7476852 0.8572727
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 17 1 5 0.8094037 0.3570602 0.7030303
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 18 2 5 0.8422477 0.3784722 0.7206061
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 19 3 5 0.862405 0.6302083 0.7483333
## maxinter: 5 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 20 5 5 0.9099148 0.7071759 0.7998485
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.8094037 0.3570602 0.7030303
## 2 2 25 0.8403277 0.6724537 0.7498485
## 3 3 25 0.8680931 0.6915509 0.7816667
## 4 5 25 0.9043775 0.6568287 0.7604545
## 5 1 20 0.8094037 0.3570602 0.7030303
## 6 2 20 0.8441743 0.6603009 0.7313636
## 7 3 20 0.8784010 0.6452546 0.7421212
## 8 5 20 0.8981848 0.6562500 0.7927273
## 9 1 15 0.8094037 0.3570602 0.7030303
## 10 2 15 0.8380603 0.6278935 0.7131818
## 11 3 15 0.8678637 0.6811343 0.7371212
## 12 5 15 0.9175098 0.6660880 0.8127273
## 13 1 10 0.8094037 0.3570602 0.7030303
## 14 2 10 0.8471101 0.6730324 0.7534848
## 15 3 10 0.8846003 0.6660880 0.7825758
## 16 5 10 0.9163630 0.7476852 0.8572727
## 17 1 5 0.8094037 0.3570602 0.7030303
## 18 2 5 0.8422477 0.3784722 0.7206061
## 19 3 5 0.8624050 0.6302083 0.7483333
## 20 5 5 0.9099148 0.7071759 0.7998485
## maxinter nodesize rocTrain rocValid rocTest
## 16 5 10 0.916363 0.7476852 0.8572727
##
## Observation 1 has a predicted value 0.293
## since this is the weighted average response across the 10 nodes it is a member of:
##
## 1) Node 110, containing 105 training observations, with node mean 0.0667 and weight 0.415 :
## irregularity <= 0.93
## 2.7 <= texture_entropy_nondir_post2
## texture_energy_nondir_post4 <= 0.0022
##
## 2) Node 6, containing 11 training observations, with node mean 0.182 and weight 0.121 :
## 480 <= texture_sumvariance_nondir_post1
## 0.58 <= max_RGH_mean
## texture_energy_nondir_post4 <= 0.0021
## texture_sumvariance_nondir_post4 <= 760
##
## 3) Node 72, containing 34 training observations, with node mean 0.794 and weight 0.111 :
## irregularity <= 0.98
## 0.0034 <= washoutRate_inside
## 0.84 <= dce3SE0
## 740 <= mean_F_r_i
##
## 4) Node 58, containing 27 training observations, with node mean 0.37 and weight 0.0935 :
## 470 <= texture_sumvariance_nondir_post1
## dce3SE0 <= 0.9
##
## 5) Node 27, containing 14 training observations, with node mean 0.5 and weight 0.0685 :
## 0.0038 <= washoutRate_inside
## 470 <= texture_sumvariance_nondir_post1
## 1.1 <= lateSE14
## irregularity <= 0.94
##
## 6) Node 106, containing 82 training observations, with node mean 0.829 and weight 0.0583 :
## 0.0038 <= washoutRate_inside
## 470 <= texture_sumvariance_nondir_post1
##
## 7) Node 83, containing 45 training observations, with node mean 0.4 and weight 0.0456 :
## irregularity <= 0.94
## 400 <= texture_sumvariance_nondir_post1
##
## 8) Node 47, containing 20 training observations, with node mean 0.05 and weight 0.0392 :
## lateSE14 <= 1.6
## 920 <= mean_F_r_i
## irregularity <= 0.92
##
## 9) Node 87, containing 50 training observations, with node mean 0.38 and weight 0.0373 :
## 390 <= texture_sumvariance_nondir_post1
## 0.42 <= earlySE9
## V6 <= 25
## texture_energy_nondir_post4 <= 0.0019
## texture_entropy_nondir_post2 <= 3.3
##
## 10) Node 121, containing 568 training observations, with node mean 0.384 and weight 0.01 :
## ROOT NODE
## id C NC pred obs
## 1 10 0.2747274 0.7252726 NC NC
## 2 69 0.3631922 0.6368078 NC C
## 3 72 0.6220447 0.3779553 C NC
## 4 81 0.5399327 0.4600673 C C
## 5 92 0.2405360 0.7594640 NC NC
## 6 95 0.2970309 0.7029691 NC C
## id C NC pred obs
## 1 10 0.2320441 0.7679559 NC NC
## 2 69 0.2687347 0.7312653 NC C
## 3 72 0.6358072 0.3641928 C NC
## 4 81 0.5371049 0.4628951 C C
## 5 92 0.3851574 0.6148426 NC NC
## 6 95 0.1906115 0.8093885 NC C
## id C NC pred obs
## 1 10 0.5697310 0.4302690 C NC
## 2 69 0.2644790 0.7355210 NC C
## 3 72 0.2167100 0.7832900 NC NC
## 4 81 0.4463127 0.5536873 NC C
## 5 92 0.2445440 0.7554560 NC NC
## 6 95 0.3804333 0.6195667 NC C
## id C NC pred obs
## 1 10 0.2934072 0.7065928 NC NC
## 2 69 0.2606203 0.7393797 NC C
## 3 72 0.3068161 0.6931839 NC NC
## 4 81 0.5579434 0.4420566 C C
## 5 92 0.3694315 0.6305685 NC NC
## 6 95 0.4991656 0.5008344 NC C
##
## Call:
## roc.default(response = perf_imgT2$obs, predictor = perf_imgT2$C)
##
## Data: perf_imgT2$C in 238 controls (perf_imgT2$obs C) > 347 cases (perf_imgT2$obs NC).
## Area under the curve: 0.7499
##
## Call:
## roc.default(response = perf_allT2$obs, predictor = perf_allT2$C)
##
## Data: perf_allT2$C in 232 controls (perf_allT2$obs C) > 353 cases (perf_allT2$obs NC).
## Area under the curve: 0.749
##
## Call:
## roc.default(response = perf_imgT1$obs, predictor = perf_imgT1$C)
##
## Data: perf_imgT1$C in 248 controls (perf_imgT1$obs C) > 337 cases (perf_imgT1$obs NC).
## Area under the curve: 0.8361
##
## Call:
## roc.default(response = perf_all$obs, predictor = perf_all$C)
##
## Data: perf_all$C in 249 controls (perf_all$obs C) > 336 cases (perf_all$obs NC).
## Area under the curve: 0.8538
## Area under the curve: 0.7499
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.36395 0.7227 0.7773 0.8319 0.5764 0.6254 0.6772
## Area under the curve: 0.749
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.4273591 0.5647 0.625 0.6853 0.728 0.7705 0.8103
## Area under the curve: 0.8361
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.3626356 0.75 0.7984 0.8468 0.6914 0.7389 0.7864
## Area under the curve: 0.8538
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.4104014 0.6827 0.739 0.7912 0.8036 0.8452 0.8839
## massB massM nonmassB nonmassM
## 209 154 126 71
## massB massM nonmassB nonmassM
## 31 14 16 6
## massB massM nonmassB nonmassM
## 209 154 126 71
## massB massM nonmassB nonmassM
## 31 14 16 6
## massB massM nonmassB nonmassM
## 209 154 126 71
## massB massM nonmassB nonmassM
## 31 14 16 6
## 0.004761905 0.05
## Selected features for group: MeanDecreaseGini imgT2
## =========NULL
## [1] "T2RGH_mean" "T2texture_entropy_nondir"
## [3] "T2grad_margin_var" "T2skew_F_r_i"
## [5] "T2texture_correlation_nondir" "T2var_F_r_i"
## [7] "ave_T210" "T2texture_inversediffmoment_nondir"
## [9] "T2max_F_r_i" "T2texture_sumaverage_nondir"
## [11] "ave_T211" "ave_T21"
## [13] "T2RGH_var" "T2texture_energy_nondir"
## [15] "ave_T217" "ave_T25"
## [17] "ave_T219" "T2mean_F_r_i"
## [19] "ave_T216" "T2texture_diffentropy_nondir"
## [21] "ave_T24" "ave_T215"
## [23] "ave_T28" "T2texture_variance_nondir"
## [25] "ave_T26" "ave_T29"
## [27] "ave_T20" "ave_T212"
## [29] "ave_T213" "ave_T218"
## [31] "T2min_F_r_i" "ave_T27"
## [33] "ave_T23" "ave_T214"
## [35] "T2_lesionSI"
## 0.02304147 0.05
## Selected features for group: MeanDecreaseGini allT2
## =========NULL
## [1] "T2RGH_mean" "ave_T210"
## [3] "T2texture_energy_nondir" "T2grad_margin"
## [5] "T2wSI_predicted" "T2max_F_r_i"
## [7] "T2var_F_r_i" "T2texture_inversediffmoment_nondir"
## [9] "ave_T20" "LMSIR_predicted"
## [11] "ave_T27" "ave_T25"
## [13] "ave_T23" "ave_T24"
## [15] "T2texture_correlation_nondir" "ave_T21"
## [17] "T2texture_variance_nondir" "ave_T213"
## [19] "ave_T26" "T2mean_F_r_i"
## [21] "ave_T22" "ave_T215"
## [23] "ave_T216" "T2min_F_r_i"
## [25] "ave_T218" "ave_T214"
## [27] "ave_T217" "T2texture_sumaverage_nondir"
## [29] "ave_T219" "ave_T28"
## [31] "T2skew_F_r_i" "T2kurt_F_r_i"
## [33] "T2texture_diffentropy_nondir" "T2_lesionSI"
## 0.08588957 0.05
## 0.03355705 0.05
## Selected features for group: MeanDecreaseGini imgT1
## =========NULL
## [1] "SER_inside" "texture_diffvariance_nondir_post1"
## [3] "texture_variance_nondir_post3" "texture_inversediffmoment_nondir_post2"
## [5] "V18" "V1"
## [7] "V7" "V6"
## [9] "texture_sumaverage_nondir_post1" "texture_diffentropy_nondir_post3"
## [11] "iiMin_change_Variance_uptake" "texture_sumentropy_nondir_post2"
## [13] "alpha_countor" "Vr_decreasingRate_inside"
## [15] "max_RGH_mean" "Slope_ini_inside"
## [17] "earlySE7" "dce3SE8"
## [19] "lateSE18" "lateSE13"
## [21] "alpha_inside" "earlySE15"
## [23] "iiiMax_Margin_Gradient" "dce3SE11"
## [25] "maxVr_countor" "earlySE19"
## [27] "dce3SE2" "dce3SE15"
## [29] "dce2SE11" "dce2SE14"
## [31] "earlySE0" "earlySE18"
## [33] "Kpeak_inside" "lateSE4"
## [35] "V5" "dce2SE19"
## [37] "Vr_decreasingRate_countor" "dce2SE17"
## [39] "texture_contrast_nondir_post4" "peakVr_countor"
## [41] "Vr_post_1_countor" "k_Max_Margin_Grad"
## [43] "skew_F_r_i" "A_inside"
## -0.1760563 0.05
## Selected features for group: MeanDecreaseGini all
## =========NULL
## [1] "texture_variance_nondir_post1" "V8"
## [3] "mean_F_r_i" "max_RGH_mean"
## [5] "texture_correlation_nondir_post3" "earlySE12"
## [7] "V6" "earlySE6"
## [9] "texture_diffentropy_nondir_post4" "kurt_F_r_i"
## [11] "T2texture_correlation_nondir" "texture_sumentropy_nondir_post2"
## [13] "Vr_post_1_inside" "edge_sharp_mean"
## [15] "ivVariance" "dce3SE19"
## [17] "earlySE17" "lateSE3"
## [19] "ave_T20" "ave_T210"
## [21] "SER_countor" "dce2SE16"
## [23] "dce2SE9" "edge_sharp_std"
## [25] "Tpeak_inside" "lateSE1"
## [27] "max_RGH_var" "ave_T26"
## [29] "A_inside"
## lesion_id cad_pt_no_txt exam_a_number_txt BIRADS lesion_label lesion_diagnosis
## 3 3 0025 7002835 4 nonmassB DENSE FIBROSIS
## 35 35 0180 4632561 4 nonmassB BENIGN BREAST TISSUE
## 36 36 0180 5254957 4 nonmassB FIBROADENOMA
## 45 45 0197 6667696 4 nonmassB LobularHyperplasia
## 46 46 0197 6667696 4 nonmassB LobularHyperplasia
## 47 47 0197 6667696 4 massB LobularHyperplasia
## 70 70 0280 5091695 4 massB BENIGN BREAST TISSUE
## 73 73 0311 6677243 4 massB FIBROADENOMA
## 75 75 0331 4722659 2 massB FIBROCYSTIC
## 76 76 0331 7347095 4 massB FIBROCYSTIC
## 77 77 0331 7347095 4 massB capillary hemangioma
## 100 100 0536 7786869 4 massB FIBROADENOMA
## 114 114 0595 7441706 4 massB BENIGN BREAST TISSUE
## 125 125 0651 4695822 4 massB FIBROADENOMA
## 130 130 0667 4864590 3 massB FIBROADENOMA
## 131 131 0667 4864590 4 massM InsituDuctal
## 132 132 0667 6993980 4 nonmassM InsituDuctal
## 142 142 0687 1201 5 massM InvasiveDuctal
## 143 143 0687 1201 5 nonmassM InvasiveDuctal
## 144 144 0687 1201 5 massM InvasiveDuctal
## 173 173 0727 4803733 4 massM InsituDuctal
## 174 174 0728 5304244 6 massB FIBROADENOMA
## 175 175 0728 5304244 4 massB DUCT PAPILLOMA WITH ATYPIA
## 176 176 0728 5304244 4 massB FIBROADENOMA
## 177 177 0728 5304244 6 nonmassM InvasiveDuctal
## 180 180 0731 5265417 4 massB DYSTROPHICCALCIFICATION
## 190 190 0745 4881779 4 massM InvasiveDuctal
## 214 214 0776 5352670 5 massB AtypicalCells
## 215 215 0776 5352670 5 massM InvasiveDuctal
## 226 226 0795 5188009 6 nonmassB ATYPICAL DUCTAL HYPERPLASIA
## 236 236 0813 5378164 5 nonmassB FIBROCYSTIC
## 237 237 0813 5378164 5 massM InvasiveLobular
## 246 246 0828 4787730 6 massM InsituDuctal
## 253 253 0839 4739257 4 massB BENIGN BREAST TISSUE
## 254 254 0839 4739257 4 massB BENIGN BREAST TISSUE
## 257 257 0844 4795902 4 nonmassM InvasiveDuctal
## 258 258 0845 5433683 5 massM InvasiveLobular
## 324 324 0954 7962026 4 massB ATYPICAL LOBULAR HYPERPLASIA
## 340 340 1008 6745959 5 massM InsituDuctal
## 341 341 1012 7629993 6 massM InvasiveDuctal
## 342 342 1012 6940724 4 massB FIBROADENOMA
## 343 343 1012 6940724 4 nonmassB FIBROADENOMA
## 365 365 1079 7417880 4 massM InsituDuctal
## 366 366 1081 7078151 4 massB FIBROADENOMA
## 389 389 2042 4964619 6 massB LobularHyperplasia
## 390 390 2042 5186978 4 massB ATYPICAL DUCTAL HYPERPLASIA
## 391 391 2042 7050570 4 massB ATYPICAL DUCTAL HYPERPLASIA
## 403 403 2071 7594721 4 massM InvasiveDuctal
## 417 417 3010 6828446 6 massM InvasiveDuctal
## 463 463 3078 4836946 5 nonmassB InsituLobular
## 470 470 3083 5345062 4 nonmassB BENIGN BREAST TISSUE
## 471 471 3086 7715466 4 nonmassB BENIGN BREAST TISSUE
## 472 472 3092 4462310 3 massB BENIGN BREAST TISSUE
## 478 478 4008 7014565 4 nonmassB ATYPICAL DUCTAL HYPERPLASIA
## 479 479 4008 7014565 6 massB HYPERPLASIA
## 491 491 4026 6998219 4 massB ATYPICAL DUCTAL HYPERPLASIA
## 494 494 4039 7041331 6 nonmassM InvasiveDuctal
## 524 524 6023 4697014 3 massB FIBROCYSTIC
## 525 525 6023 4697014 3 massB SCLEROSING ADENOSIS
## 531 531 6026 4888386 4 nonmassM InsituDuctal
## 557 557 6043 5249778 4 nonmassB ATYPICAL DUCTAL HYPERPLASIA
## 576 576 6069 7581124 4 massB BENIGN BREAST TISSUE
## 577 577 6069 7581124 4 massB BENIGN BREAST TISSUE
## 587 587 6164 6971531 4 massB BENIGN BREAST TISSUE
## 588 588 6174 7009629 4 nonmassB ATYPICAL DUCTAL HYPERPLASIA
## 633 633 7193 7347138 4 nonmassB BENIGN BREAST TISSUE
## 634 634 7193 7347138 4 nonmassB BENIGN BREAST TISSUE
## find_t2_signal_int
## 3 None
## 35 Hypointense or not seen
## 36 Hypointense or not seen
## 45 Hypointense or not seen
## 46 Hypointense or not seen
## 47 Slightly hyperintense
## 70 None
## 73 None
## 75 None
## 76 Hyperintense
## 77 None
## 100 Hypointense or not seen
## 114 Hypointense or not seen
## 125 Hyperintense
## 130 Hypointense or not seen
## 131 Hyperintense
## 132 Hyperintense
## 142 Slightly hyperintense
## 143 Slightly hyperintense
## 144 Slightly hyperintense
## 173 Slightly hyperintense
## 174 Slightly hyperintense
## 175 Hyperintense
## 176 None
## 177 None
## 180 Slightly hyperintense
## 190 None
## 214 Hypointense or not seen
## 215 None
## 226 Hypointense or not seen
## 236 None
## 237 None
## 246 None
## 253 None
## 254 None
## 257 None
## 258 None
## 324 None
## 340 None
## 341 None
## 342 Hypointense or not seen
## 343 Hyperintense
## 365 None
## 366 Hyperintense
## 389 Hyperintense
## 390 Hypointense or not seen
## 391 None
## 403 Slightly hyperintense
## 417 None
## 463 None
## 470 None
## 471 None
## 472 Hypointense or not seen
## 478 None
## 479 None
## 491 Hypointense or not seen
## 494 None
## 524 None
## 525 None
## 531 None
## 557 None
## 576 Hyperintense
## 577 Hypointense or not seen
## 587 Hypointense or not seen
## 588 None
## 633 None
## 634 Hypointense or not seen
##
## ============ bagging trees treedata_imgT2
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6839138 0.4356383 0.402849
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.7459834 0.6164894 0.654416
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.7925572 0.7271277 0.717094
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 4 1 20 0.6839138 0.4356383 0.402849
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 2 20 0.7546202 0.6425532 0.682906
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 3 20 0.7900896 0.6164894 0.6891738
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 7 1 15 0.6839138 0.4356383 0.402849
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 8 2 15 0.744796 0.5973404 0.6444444
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 3 15 0.7974395 0.6617021 0.702849
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 10 1 10 0.6839138 0.4356383 0.402849
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 11 2 10 0.740126 0.587766 0.6363248
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 12 3 10 0.8061161 0.6494681 0.6849003
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 5 0.6839138 0.4414894 0.4081197
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 5 0.7454992 0.5329787 0.6108262
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 5 0.7919071 0.4218085 0.6461538
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6839138 0.4356383 0.4028490
## 2 2 25 0.7459834 0.6164894 0.6544160
## 3 3 25 0.7925572 0.7271277 0.7170940
## 4 1 20 0.6839138 0.4356383 0.4028490
## 5 2 20 0.7546202 0.6425532 0.6829060
## 6 3 20 0.7900896 0.6164894 0.6891738
## 7 1 15 0.6839138 0.4356383 0.4028490
## 8 2 15 0.7447960 0.5973404 0.6444444
## 9 3 15 0.7974395 0.6617021 0.7028490
## 10 1 10 0.6839138 0.4356383 0.4028490
## 11 2 10 0.7401260 0.5877660 0.6363248
## 12 3 10 0.8061161 0.6494681 0.6849003
## 13 1 5 0.6839138 0.4414894 0.4081197
## 14 2 5 0.7454992 0.5329787 0.6108262
## 15 3 5 0.7919071 0.4218085 0.6461538
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.7925572 0.7271277 0.717094
##
## Observation 1 has a predicted value 0.623
## since this is the weighted average response across the 8 nodes it is a member of:
##
## 1) Node 16, containing 32 training observations, with node mean 0.688 and weight 0.277 :
## 3.5 <= T2texture_entropy_nondir
## ave_T219 <= 140
##
## 2) Node 20, containing 37 training observations, with node mean 0.526 and weight 0.212 :
## 280 <= T2RGH_var
## 41 <= T2RGH_mean
## T2texture_energy_nondir <= 0.00072
##
## 3) Node 35, containing 79 training observations, with node mean 0.684 and weight 0.166 :
## T2texture_energy_nondir <= 0.00064
## 160 <= ave_T210
## T2grad_margin_var <= 7200
##
## 4) Node 26, containing 52 training observations, with node mean 0.596 and weight 0.14 :
## 22 <= T2RGH_mean
## T2texture_correlation_nondir <= 0.26
## 3.5 <= T2texture_entropy_nondir
##
## 5) Node 40, containing 100 training observations, with node mean 0.65 and weight 0.116 :
## 3.4 <= T2texture_entropy_nondir
## 0.071 <= T2texture_inversediffmoment_nondir
## ave_T214 <= 320
##
## 6) Node 42, containing 103 training observations, with node mean 0.635 and weight 0.0454 :
## 22 <= T2RGH_mean
## T2texture_energy_nondir <= 0.00054
## ave_T214 <= 310
##
## 7) Node 39, containing 92 training observations, with node mean 0.474 and weight 0.034 :
## 59 <= ave_T210
## T2texture_correlation_nondir <= 0.26
## 510 <= T2RGH_var
##
## 8) Node 54, containing 557 training observations, with node mean 0.402 and weight 0.01 :
## ROOT NODE
##
## ============ bagging trees treedata_allT2
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6951045 0.662234 0.6399597
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.7486235 0.637234 0.6728999
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.7958474 0.6430851 0.7041139
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 4 1 20 0.6951045 0.662234 0.6399597
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 2 20 0.7417313 0.6393617 0.6634062
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 3 20 0.8040398 0.6351064 0.7252589
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 7 1 15 0.6951045 0.662234 0.6399597
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 8 2 15 0.7546269 0.6531915 0.6697353
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 3 15 0.814335 0.6351064 0.7205121
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 10 1 10 0.6951045 0.662234 0.6399597
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 11 2 10 0.7507463 0.631383 0.6839758
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 12 3 10 0.7867993 0.6611702 0.7138953
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 5 0.6951045 0.662234 0.6399597
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 5 0.7487363 0.6398936 0.6790852
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 5 0.7990182 0.6223404 0.691168
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6951045 0.6622340 0.6399597
## 2 2 25 0.7486235 0.6372340 0.6728999
## 3 3 25 0.7958474 0.6430851 0.7041139
## 4 1 20 0.6951045 0.6622340 0.6399597
## 5 2 20 0.7417313 0.6393617 0.6634062
## 6 3 20 0.8040398 0.6351064 0.7252589
## 7 1 15 0.6951045 0.6622340 0.6399597
## 8 2 15 0.7546269 0.6531915 0.6697353
## 9 3 15 0.8143350 0.6351064 0.7205121
## 10 1 10 0.6951045 0.6622340 0.6399597
## 11 2 10 0.7507463 0.6313830 0.6839758
## 12 3 10 0.7867993 0.6611702 0.7138953
## 13 1 5 0.6951045 0.6622340 0.6399597
## 14 2 5 0.7487363 0.6398936 0.6790852
## 15 3 5 0.7990182 0.6223404 0.6911680
## maxinter nodesize rocTrain rocValid rocTest
## 6 3 20 0.8040398 0.6351064 0.7252589
##
## Observation 1 has a predicted value 0.593
## since this is the weighted average response across the 6 nodes it is a member of:
##
## 1) Node 34, containing 0.5 training observations, with node mean 0.497 and weight 0.277 :
## T2wSI_predicted = None
## T2var_F_r_i <= 8600
## 60 <= ave_T210
##
## 2) Node 36, containing 0.5 training observations, with node mean 0.632 and weight 0.255 :
## T2wSI_predicted = None
## 370 <= T2max_F_r_i
## T2kurt_F_r_i <= 4.6
##
## 3) Node 27, containing 78 training observations, with node mean 0.654 and weight 0.164 :
## T2texture_energy_nondir <= 0.00054
## 170 <= ave_T210
##
## 4) Node 32, containing 0.5 training observations, with node mean 0.447 and weight 0.15 :
## T2wSI_predicted = None
## T2skew_F_r_i <= 1.3
## 0.095 <= T2texture_correlation_nondir
##
## 5) Node 19, containing 41 training observations, with node mean 0.805 and weight 0.143 :
## T2texture_energy_nondir <= 0.00048
## 170 <= ave_T210 <= 280
##
## 6) Node 41, containing 557 training observations, with node mean 0.402 and weight 0.01 :
## ROOT NODE
##
## ============ bagging trees treedata_imgT1
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7615987 0.737234 0.7608262
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.8116219 0.7558511 0.8206553
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.867602 0.7829787 0.8283476
## maxinter: 5 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 4 5 25 0.9031841 0.8117021 0.862963
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 1 20 0.7675954 0.7457447 0.7698006
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 2 20 0.8122852 0.7420213 0.7964387
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 7 3 20 0.8515423 0.7744681 0.8219373
## maxinter: 5 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 8 5 20 0.8947463 0.8117021 0.842735
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 1 15 0.7675954 0.7457447 0.7698006
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 10 2 15 0.8084245 0.7808511 0.8173789
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 11 3 15 0.8601857 0.7914894 0.8391738
## maxinter: 5 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 12 5 15 0.8943483 0.7978723 0.8467236
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 10 0.7660697 0.7531915 0.7696581
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 10 0.8143682 0.7382979 0.7954416
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 10 0.8631642 0.7765957 0.8246439
## maxinter: 5 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 16 5 10 0.8924245 0.8074468 0.8438746
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 17 1 5 0.7660697 0.7531915 0.7696581
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 18 2 5 0.8167828 0.7771277 0.8264957
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 19 3 5 0.8532007 0.7829787 0.8236467
## maxinter: 5 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 20 5 5 0.8971808 0.7946809 0.8592593
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7615987 0.7372340 0.7608262
## 2 2 25 0.8116219 0.7558511 0.8206553
## 3 3 25 0.8676020 0.7829787 0.8283476
## 4 5 25 0.9031841 0.8117021 0.8629630
## 5 1 20 0.7675954 0.7457447 0.7698006
## 6 2 20 0.8122852 0.7420213 0.7964387
## 7 3 20 0.8515423 0.7744681 0.8219373
## 8 5 20 0.8947463 0.8117021 0.8427350
## 9 1 15 0.7675954 0.7457447 0.7698006
## 10 2 15 0.8084245 0.7808511 0.8173789
## 11 3 15 0.8601857 0.7914894 0.8391738
## 12 5 15 0.8943483 0.7978723 0.8467236
## 13 1 10 0.7660697 0.7531915 0.7696581
## 14 2 10 0.8143682 0.7382979 0.7954416
## 15 3 10 0.8631642 0.7765957 0.8246439
## 16 5 10 0.8924245 0.8074468 0.8438746
## 17 1 5 0.7660697 0.7531915 0.7696581
## 18 2 5 0.8167828 0.7771277 0.8264957
## 19 3 5 0.8532007 0.7829787 0.8236467
## 20 5 5 0.8971808 0.7946809 0.8592593
## maxinter nodesize rocTrain rocValid rocTest
## 4 5 25 0.9031841 0.8117021 0.862963
##
## Observation 1 has a predicted value 0.173
## since this is the weighted average response across the 9 nodes it is a member of:
##
## 1) Node 100, containing 135 training observations, with node mean 0.133 and weight 0.211 :
## texture_diffvariance_nondir_post1 <= 140
## Vr_post_1_countor <= 0.14
## 1.6 <= A_inside
##
## 2) Node 51, containing 33 training observations, with node mean 0.273 and weight 0.173 :
## SER_inside <= 0.81
## texture_diffvariance_nondir_post1 <= 110
## Vr_post_1_countor <= 0.11
## 20 <= V6
##
## 3) Node 101, containing 140 training observations, with node mean 0.113 and weight 0.157 :
## SER_inside <= 0.7
## lateSE18 <= 1.9
## 0.52 <= max_RGH_mean
## V18 <= 32
##
## 4) Node 103, containing 275 training observations, with node mean 0.211 and weight 0.124 :
## texture_diffvariance_nondir_post1 <= 110
## SER_inside <= 0.8
##
## 5) Node 64, containing 42 training observations, with node mean 0.19 and weight 0.0968 :
## Slope_ini_inside <= 0.66
## SER_inside <= 0.67
## skew_F_r_i <= 0.75
## 18 <= V1
##
## 6) Node 94, containing 82 training observations, with node mean 0.11 and weight 0.0952 :
## Slope_ini_inside <= 0.85
## V5 <= 16
## SER_inside <= 0.67
##
## 7) Node 63, containing 41 training observations, with node mean 0.195 and weight 0.0756 :
## SER_inside <= 0.8
## texture_diffvariance_nondir_post1 <= 140
## 15 <= V18
## 0.019 <= Vr_decreasingRate_inside
##
## 8) Node 83, containing 70 training observations, with node mean 0.1 and weight 0.057 :
## texture_diffvariance_nondir_post1 <= 100
## SER_inside <= 0.82
## Vr_post_1_countor <= 0.14
## 1.1 <= lateSE4
##
## 9) Node 104, containing 560 training observations, with node mean 0.402 and weight 0.01 :
## ROOT NODE
##
## ============ bagging trees treedata_all
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7384013 0.7393617 0.7256024
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.8076551 0.7765957 0.7850904
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8645572 0.8297872 0.840512
## maxinter: 5 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 4 5 25 0.89799 0.8074468 0.8340361
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 1 20 0.7384013 0.7393617 0.7256024
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 2 20 0.8166302 0.7803191 0.7986446
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 7 3 20 0.8596949 0.775 0.8197289
## maxinter: 5 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 8 5 20 0.9117944 0.8265957 0.889759
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 1 15 0.7384013 0.7393617 0.7256024
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 10 2 15 0.8095191 0.7771277 0.785994
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 11 3 15 0.8550779 0.7946809 0.8120482
## maxinter: 5 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 12 5 15 0.9079138 0.8340426 0.8665663
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 10 0.7384013 0.7393617 0.7256024
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 10 0.8136451 0.8005319 0.7971386
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 10 0.8589187 0.781383 0.8176205
## maxinter: 5 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 16 5 10 0.9096982 0.7680851 0.8403614
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 17 1 5 0.7384013 0.7393617 0.7256024
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 18 2 5 0.8235821 0.8074468 0.810994
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 19 3 5 0.8642521 0.8414894 0.8549699
## maxinter: 5 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 20 5 5 0.9064146 0.7968085 0.851506
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7384013 0.7393617 0.7256024
## 2 2 25 0.8076551 0.7765957 0.7850904
## 3 3 25 0.8645572 0.8297872 0.8405120
## 4 5 25 0.8979900 0.8074468 0.8340361
## 5 1 20 0.7384013 0.7393617 0.7256024
## 6 2 20 0.8166302 0.7803191 0.7986446
## 7 3 20 0.8596949 0.7750000 0.8197289
## 8 5 20 0.9117944 0.8265957 0.8897590
## 9 1 15 0.7384013 0.7393617 0.7256024
## 10 2 15 0.8095191 0.7771277 0.7859940
## 11 3 15 0.8550779 0.7946809 0.8120482
## 12 5 15 0.9079138 0.8340426 0.8665663
## 13 1 10 0.7384013 0.7393617 0.7256024
## 14 2 10 0.8136451 0.8005319 0.7971386
## 15 3 10 0.8589187 0.7813830 0.8176205
## 16 5 10 0.9096982 0.7680851 0.8403614
## 17 1 5 0.7384013 0.7393617 0.7256024
## 18 2 5 0.8235821 0.8074468 0.8109940
## 19 3 5 0.8642521 0.8414894 0.8549699
## 20 5 5 0.9064146 0.7968085 0.8515060
## maxinter nodesize rocTrain rocValid rocTest
## 8 5 20 0.9117944 0.8265957 0.889759
##
## Observation 1 has a predicted value 0.11
## since this is the weighted average response across the 9 nodes it is a member of:
##
## 1) Node 87, containing 74 training observations, with node mean 0.0676 and weight 0.343 :
## texture_variance_nondir_post1 <= 79
## SER_countor <= 0.71
## ivVariance <= 0.013
## texture_sumentropy_nondir_post2 <= 1.8
##
## 2) Node 95, containing 109 training observations, with node mean 0.127 and weight 0.296 :
## 6 <= Tpeak_inside
## texture_correlation_nondir_post3 <= 0.31
## mean_F_r_i <= 920
## 110 <= ave_T210
##
## 3) Node 49, containing 25 training observations, with node mean 0.04 and weight 0.127 :
## SER_countor <= 0.7
## 20 <= V6
## texture_variance_nondir_post1 <= 45
##
## 4) Node 82, containing 63 training observations, with node mean 0.175 and weight 0.115 :
## texture_variance_nondir_post1 <= 170
## 0.61 <= lateSE3
## texture_diffentropy_nondir_post4 <= 1.3
## ivVariance <= 0.02
## 1.6 <= A_inside
##
## 5) Node 30, containing 16 training observations, with node mean 0.0625 and weight 0.0362 :
## 6.1 <= Tpeak_inside
## texture_variance_nondir_post1 <= 190
## 19 <= V6
## max_RGH_var <= 0.09
## ivVariance <= 0.0077
##
## 6) Node 91, containing 87 training observations, with node mean 0.092 and weight 0.0293 :
## texture_variance_nondir_post1 <= 79
## SER_countor <= 0.71
## ivVariance <= 0.013
##
## 7) Node 64, containing 39 training observations, with node mean 0.385 and weight 0.0255 :
## texture_variance_nondir_post1 <= 190
## 6.6 <= Tpeak_inside
## mean_F_r_i <= 940
## texture_correlation_nondir_post3 <= 0.44
## max_RGH_mean <= 0.53
##
## 8) Node 51, containing 27 training observations, with node mean 0.296 and weight 0.0186 :
## earlySE12 <= 0.85
## texture_variance_nondir_post1 <= 190
## max_RGH_mean <= 0.53
## 1.6 <= A_inside
##
## 9) Node 100, containing 560 training observations, with node mean 0.402 and weight 0.01 :
## ROOT NODE
## id C NC pred obs
## 1 3 0.6230303 0.3769697 C NC
## 2 35 0.3529939 0.6470061 NC NC
## 3 36 0.4151224 0.5848776 NC NC
## 4 45 0.3017016 0.6982984 NC NC
## 5 46 0.3456295 0.6543705 NC NC
## 6 47 0.3481065 0.6518935 NC NC
## id C NC pred obs
## 1 3 0.5927406 0.4072594 C NC
## 2 35 0.2069604 0.7930396 NC NC
## 3 36 0.3869609 0.6130391 NC NC
## 4 45 0.2088778 0.7911222 NC NC
## 5 46 0.2088778 0.7911222 NC NC
## 6 47 0.3180411 0.6819589 NC NC
## id C NC pred obs
## 1 3 0.1726708 0.8273292 NC NC
## 2 35 0.2088135 0.7911865 NC NC
## 3 36 0.1695515 0.8304485 NC NC
## 4 45 0.1749261 0.8250739 NC NC
## 5 46 0.1593673 0.8406327 NC NC
## 6 47 0.2568477 0.7431523 NC NC
## id C NC pred obs
## 1 3 0.1102743 0.8897257 NC NC
## 2 35 0.1089283 0.8910717 NC NC
## 3 36 0.2701998 0.7298002 NC NC
## 4 45 0.1960654 0.8039346 NC NC
## 5 46 0.1489373 0.8510627 NC NC
## 6 47 0.2594300 0.7405700 NC NC
##
## Call:
## roc.default(response = perf_imgT2$obs, predictor = perf_imgT2$C)
##
## Data: perf_imgT2$C in 283 controls (perf_imgT2$obs C) > 425 cases (perf_imgT2$obs NC).
## Area under the curve: 0.7435
##
## Call:
## roc.default(response = perf_allT2$obs, predictor = perf_allT2$C)
##
## Data: perf_allT2$C in 276 controls (perf_allT2$obs C) > 432 cases (perf_allT2$obs NC).
## Area under the curve: 0.7399
##
## Call:
## roc.default(response = perf_imgT1$obs, predictor = perf_imgT1$C)
##
## Data: perf_imgT1$C in 293 controls (perf_imgT1$obs C) > 415 cases (perf_imgT1$obs NC).
## Area under the curve: 0.8408
##
## Call:
## roc.default(response = perf_all$obs, predictor = perf_all$C)
##
## Data: perf_all$C in 289 controls (perf_all$obs C) > 419 cases (perf_all$obs NC).
## Area under the curve: 0.8586
## Area under the curve: 0.7435
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.3638613 0.7208 0.7703 0.8198 0.5624 0.6118 0.6588
## Area under the curve: 0.7399
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.432035 0.5543 0.6123 0.6667 0.7245 0.7639 0.8032
## Area under the curve: 0.8408
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.4158285 0.6724 0.7235 0.7713 0.7831 0.8217 0.8578
## Area under the curve: 0.8586
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.4104014 0.6955 0.7474 0.7958 0.8043 0.8425 0.8783
## massB massM nonmassB nonmassM
## 222 149 121 72
## massB massM nonmassB nonmassM
## 18 19 21 5
## massB massM nonmassB nonmassM
## 222 149 121 72
## massB massM nonmassB nonmassM
## 18 19 21 5
## massB massM nonmassB nonmassM
## 222 149 121 72
## massB massM nonmassB nonmassM
## 18 19 21 5
## 0.03902439 0.05
## Selected features for group: MeanDecreaseGini imgT2
## =========NULL
## [1] "T2RGH_var" "T2kurt_F_r_i"
## [3] "T2texture_inversediffmoment_nondir" "T2grad_margin_var"
## [5] "ave_T210" "ave_T27"
## [7] "T2texture_entropy_nondir" "T2max_F_r_i"
## [9] "T2var_F_r_i" "T2texture_correlation_nondir"
## [11] "T2skew_F_r_i" "T2RGH_mean"
## [13] "ave_T20" "ave_T21"
## [15] "ave_T212" "ave_T216"
## [17] "ave_T24" "T2texture_sumvariance_nondir"
## [19] "T2texture_diffvariance_nondir" "T2texture_diffentropy_nondir"
## [21] "T2grad_margin" "ave_T26"
## [23] "ave_T22" "ave_T29"
## [25] "ave_T218" "T2texture_energy_nondir"
## [27] "ave_T217" "ave_T23"
## [29] "ave_T211" "T2mean_F_r_i"
## [31] "ave_T25" "T2min_F_r_i"
## [33] "ave_T213" "ave_T28"
## [35] "T2_lesionSI"
## 0 0.05
## Selected features for group: MeanDecreaseGini allT2
## =========NULL
## [1] "T2RGH_var" "T2RGH_mean"
## [3] "T2skew_F_r_i" "T2wSI_predicted"
## [5] "ave_T210" "ave_T28"
## [7] "T2texture_contrast_nondir" "ave_T213"
## [9] "T2grad_margin_var" "T2texture_energy_nondir"
## [11] "ave_T211" "ave_T218"
## [13] "ave_T24" "T2texture_sumentropy_nondir"
## [15] "LMSIR_predicted" "ave_T22"
## [17] "ave_T212" "ave_T216"
## [19] "ave_T29" "T2_lesionSI"
## [21] "ave_T26" "T2min_F_r_i"
## [23] "ave_T219" "ave_T20"
## [25] "ave_T217" "ave_T25"
## [27] "T2texture_correlation_nondir" "T2kurt_F_r_i"
## [29] "ave_T215" "T2texture_sumvariance_nondir"
## [31] "T2_lesionSIstd"
## 0.08387097 0.05
## -0.0915493 0.05
## Selected features for group: MeanDecreaseGini imgT1
## =========NULL
## [1] "texture_sumvariance_nondir_post2" "circularity"
## [3] "earlySE10" "V1"
## [5] "min_F_r_i" "texture_energy_nondir_post3"
## [7] "V13" "texture_correlation_nondir_post1"
## [9] "Kpeak_countor" "V12"
## [11] "kurt_F_r_i" "texture_sumentropy_nondir_post3"
## [13] "V14" "maxVr_countor"
## [15] "Vr_decreasingRate_inside" "dce3SE2"
## [17] "texture_energy_nondir_post1" "dce2SE1"
## [19] "lateSE9" "lateSE1"
## [21] "dce2SE4" "texture_variance_nondir_post3"
## [23] "max_RGH_mean" "iAUC1_inside"
## [25] "V2" "SER_inside"
## [27] "lateSE3" "A_countor"
## [29] "iMax_Variance_uptake" "washoutRate_inside"
## [31] "peakCr_countor" "alpha_inside"
## [33] "V4" "dce3SE3"
## [35] "texture_sumvariance_nondir_post4" "max_RGH_var"
## [37] "A_inside"
## 0.01910828 0.05
## Selected features for group: MeanDecreaseGini all
## =========NULL
## [1] "texture_sumvariance_nondir_post2" "texture_sumvariance_nondir_post1"
## [3] "Slope_ini_countor" "max_RGH_mean"
## [5] "alpha_inside" "V14"
## [7] "V6" "V13"
## [9] "max_F_r_i" "T2texture_correlation_nondir"
## [11] "beta_inside" "texture_sumaverage_nondir_post4"
## [13] "iiiMax_Margin_Gradient" "texture_diffentropy_nondir_post3"
## [15] "skew_F_r_i" "texture_entropy_nondir_post3"
## [17] "dce3SE15" "dce3SE4"
## [19] "ave_T27" "ave_T210"
## [21] "texture_energy_nondir_post4" "ave_T213"
## [23] "texture_inversediffmoment_nondir_post1" "texture_entropy_nondir_post2"
## [25] "ave_T216" "T2texture_diffentropy_nondir"
## [27] "ave_T22" "ave_T217"
## [29] "T2min_F_r_i" "lateSE12"
## [31] "Vr_increasingRate_countor" "lateSE10"
## [33] "A_countor" "Vr_increasingRate_inside"
## [35] "V17" "A_inside"
## lesion_id cad_pt_no_txt exam_a_number_txt BIRADS lesion_label lesion_diagnosis
## 15 15 0122 5108281 3 massB Cyst
## 16 16 0123 6909758 4 nonmassB COLUMNAR CELL CHANGES
## 17 17 0123 6909758 4 nonmassB BENIGN BREAST TISSUE
## 39 39 0103 6836585 5 massM PHYLLODES TUMOR
## 42 42 0196 5289117 4 nonmassB SCLEROSING ADENOSIS
## 43 43 0196 5289117 4 nonmassB ColumnarAlterationwoAtypia
## 44 44 0196 5289117 4 nonmassB ColumnarAlterationwoAtypia
## 50 50 0199 4362726 4 massB ATYPICAL LOBULAR HYPERPLASIA
## 78 78 0352 4785776 4 massB FIBROADENOMA
## 79 79 0357 5137030 4 nonmassB FIBROCYSTIC
## 80 80 0376 4609403 4 massB BENIGN HAMARTOMA
## 88 88 0456 6689214 4 massM InvasiveDuctal
## 113 113 0586 5332925 4 nonmassM InsituDuctal
## 137 137 0681 4999374 3 massB FIBROCYSTIC
## 154 154 0705 4648471 5 massM InvasiveDuctal
## 162 162 0714 5324209 5 massM InvasiveDuctal
## 163 163 0714 5324209 5 massM InvasiveDuctal
## 164 164 0714 5324209 5 nonmassM InsituDuctal
## 166 166 0720 4965525 4 massB FIBROCYSTIC
## 167 167 0720 4965525 4 nonmassB FIBROADENOMA
## 169 169 0722 5366177 5 massM InvasiveDuctal
## 183 183 0736 4963473 4 massB BENIGN BREAST TISSUE
## 192 192 0752 4940477 4 nonmassB BENIGN BREAST TISSUE
## 193 193 0752 4940477 4 nonmassB FIBROCYSTIC
## 194 194 0752 4940477 4 nonmassB BENIGN BREAST TISSUE
## 203 203 0771 4680997 4 massB DUCT PAPILLOMA
## 204 204 0771 4680997 4 massB DUCT PAPILLOMA
## 208 208 0778 4794199 5 massB FIBROADENOMA
## 251 251 0834 4614262 5 massM InvasiveDuctal
## 260 260 0847 5064132 4 massM InvasiveDuctal
## 291 291 0876 4719378 4 massB BENIGN BREAST TISSUE
## 328 328 0967 6938015 4 nonmassB BENIGN BREAST TISSUE
## 329 329 0967 6938015 4 nonmassM InvasiveDuctal
## 346 346 1024 6980462 4 nonmassB ATYPICAL DUCTAL HYPERPLASIA
## 347 347 1024 6980462 4 nonmassB DUCT PAPILLOMA WITH ATYPIA
## 348 348 1025 6703528 4 nonmassB FIBROCYSTIC
## 350 350 1044 7366817 5 massM InvasiveDuctal
## 355 355 1062 7408296 4 massB ATYPICAL LOBULAR HYPERPLASIA
## 356 356 1062 7408296 4 nonmassB FIBROCYSTIC
## 362 362 1077 6890028 4 massB BENIGN BREAST TISSUE
## 379 379 2017 7397047 6 massB RADIAL SCLEROSING LESION
## 386 386 2033 6849696 6 massM InvasiveLobular
## 387 387 2033 6849696 4 nonmassB ATYPICAL LOBULAR HYPERPLASIA
## 394 394 2050 6689745 6 massM InvasiveDuctal
## 429 429 3031 7106716 3 massB COLUMNAR CELL CHANGES
## 433 433 3039 6894870 4 massB HYPERPLASIA
## 483 483 4018 6983262 6 nonmassM InsituDuctal
## 484 484 4019 7151338 4 massB BENIGN BREAST TISSUE
## 486 486 4021 6992707 4 nonmassB ADENOSIS
## 532 532 6027 4770166 4 massB BENIGN BREAST TISSUE
## 533 533 6027 4770166 4 nonmassB BENIGN BREAST TISSUE
## 541 541 6035 5062962 5 massM InvasiveDuctal
## 542 542 6035 5062962 5 massM InvasiveDuctal
## 543 543 6035 5062962 5 massM InvasiveDuctal
## 560 560 6045 5208117 6 massM InvasiveDuctal
## 561 561 6045 5208117 6 massM InvasiveDuctal
## 567 567 6048 5284266 6 massM InvasiveDuctal
## 579 579 6101 5087078 4 nonmassB FIBROCYSTIC
## 580 580 6101 7709238 6 massM InvasiveDuctal
## 606 606 7076 7267446 3 nonmassM InsituDuctal
## 607 607 7076 7267446 3 nonmassB ATYPICAL DUCTAL HYPERPLASIA
## 628 628 7186 5263507 6 massM InvasiveDuctal
## 632 632 7192 7974056 4 nonmassB BENIGN BREAST TISSUE
## find_t2_signal_int
## 15 None
## 16 None
## 17 None
## 39 None
## 42 Hypointense or not seen
## 43 None
## 44 Hypointense or not seen
## 50 Hyperintense
## 78 Slightly hyperintense
## 79 Hypointense or not seen
## 80 Slightly hyperintense
## 88 None
## 113 Hypointense or not seen
## 137 Slightly hyperintense
## 154 Slightly hyperintense
## 162 None
## 163 None
## 164 None
## 166 Slightly hyperintense
## 167 Hypointense or not seen
## 169 Hyperintense
## 183 Slightly hyperintense
## 192 None
## 193 None
## 194 None
## 203 None
## 204 None
## 208 Hypointense or not seen
## 251 None
## 260 Hyperintense
## 291 None
## 328 None
## 329 None
## 346 None
## 347 None
## 348 Hypointense or not seen
## 350 None
## 355 Hypointense or not seen
## 356 Hypointense or not seen
## 362 Slightly hyperintense
## 379 None
## 386 None
## 387 None
## 394 None
## 429 Hypointense or not seen
## 433 None
## 483 None
## 484 Hyperintense
## 486 Hypointense or not seen
## 532 None
## 533 None
## 541 None
## 542 None
## 543 None
## 560 Hypointense or not seen
## 561 Hypointense or not seen
## 567 None
## 579 None
## 580 Hypointense or not seen
## 606 None
## 607 None
## 628 None
## 632 None
##
## ============ bagging trees treedata_imgT2
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6676121 0.732906 0.7317187
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.7529253 0.7510684 0.7478125
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8192222 0.7035256 0.7776562
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 4 1 20 0.6676121 0.732906 0.7317187
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 2 20 0.7436777 0.7761752 0.7778125
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 3 20 0.7978444 0.7008547 0.75125
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 7 1 15 0.6676121 0.732906 0.7317187
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 8 2 15 0.7352875 0.786859 0.7732812
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 3 15 0.8083123 0.6741453 0.7376563
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 10 1 10 0.6676121 0.732906 0.7317187
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 11 2 10 0.7518832 0.7761752 0.7771875
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 12 3 10 0.8113795 0.7574786 0.7904688
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 5 0.6676121 0.732906 0.7317187
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 5 0.7581692 0.7804487 0.7835937
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 5 0.8006016 0.6538462 0.7148438
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6676121 0.7329060 0.7317187
## 2 2 25 0.7529253 0.7510684 0.7478125
## 3 3 25 0.8192222 0.7035256 0.7776562
## 4 1 20 0.6676121 0.7329060 0.7317187
## 5 2 20 0.7436777 0.7761752 0.7778125
## 6 3 20 0.7978444 0.7008547 0.7512500
## 7 1 15 0.6676121 0.7329060 0.7317187
## 8 2 15 0.7352875 0.7868590 0.7732812
## 9 3 15 0.8083123 0.6741453 0.7376563
## 10 1 10 0.6676121 0.7329060 0.7317187
## 11 2 10 0.7518832 0.7761752 0.7771875
## 12 3 10 0.8113795 0.7574786 0.7904688
## 13 1 5 0.6676121 0.7329060 0.7317187
## 14 2 5 0.7581692 0.7804487 0.7835937
## 15 3 5 0.8006016 0.6538462 0.7148438
## maxinter nodesize rocTrain rocValid rocTest
## 12 3 10 0.8113795 0.7574786 0.7904688
##
## Observation 1 has a predicted value 0.311
## since this is the weighted average response across the 12 nodes it is a member of:
##
## 1) Node 48, containing 101 training observations, with node mean 0.194 and weight 0.205 :
## 8.5 <= T2min_F_r_i
## ave_T210 <= 300
##
## 2) Node 50, containing 121 training observations, with node mean 0.397 and weight 0.121 :
## T2texture_entropy_nondir <= 3.4
## 0.28 <= T2texture_correlation_nondir
##
## 3) Node 44, containing 83 training observations, with node mean 0.289 and weight 0.118 :
## T2texture_entropy_nondir <= 3.4
## 0.25 <= T2texture_correlation_nondir
## 110 <= T2mean_F_r_i
##
## 4) Node 51, containing 126 training observations, with node mean 0.294 and weight 0.118 :
## T2RGH_var <= 350
## T2texture_inversediffmoment_nondir <= 0.24
## 320 <= T2texture_sumvariance_nondir
##
## 5) Node 49, containing 104 training observations, with node mean 0.202 and weight 0.114 :
## T2RGH_mean <= 24
## -0.85 <= T2kurt_F_r_i
## T2grad_margin_var <= 2700
##
## 6) Node 53, containing 140 training observations, with node mean 0.586 and weight 0.107 :
## 0.26 <= T2texture_correlation_nondir
## T2grad_margin <= 57
## 42 <= T2texture_diffvariance_nondir
##
## 7) Node 62, containing 261 training observations, with node mean 0.35 and weight 0.0854 :
## T2skew_F_r_i <= 1.3
## 68 <= ave_T210
## T2texture_entropy_nondir <= 3.5
##
## 8) Node 41, containing 72 training observations, with node mean 0.194 and weight 0.0555 :
## T2texture_entropy_nondir <= 3.4
## 58 <= T2mean_F_r_i
## 26 <= T2min_F_r_i
##
## 9) Node 65, containing 386 training observations, with node mean 0.389 and weight 0.0436 :
## T2texture_entropy_nondir <= 3.5
## 39 <= ave_T210
## T2texture_diffentropy_nondir <= 1.8
##
## 10) Node 64, containing 340 training observations, with node mean 0.374 and weight 0.013 :
## T2texture_entropy_nondir <= 3.4
## 41 <= ave_T210
## T2texture_sumvariance_nondir <= 1600
##
## 11) Node 66, containing 562 training observations, with node mean 0.392 and weight 0.01 :
## ROOT NODE
##
## 12) Node 59, containing 216 training observations, with node mean 0.259 and weight 0.0097 :
## T2RGH_var <= 350
## T2texture_inversediffmoment_nondir <= 0.24
##
## ============ bagging trees treedata_allT2
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6818464 0.6816239 0.6971852
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.7466855 0.6762821 0.7271111
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8039853 0.6388889 0.7151111
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 4 1 20 0.6818464 0.6816239 0.6971852
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 2 20 0.7338232 0.6976496 0.7420741
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 3 20 0.8136485 0.588141 0.7131852
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 7 1 15 0.6818464 0.6816239 0.6971852
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 8 2 15 0.7436249 0.7035256 0.7508148
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 3 15 0.8090775 0.5480769 0.7251852
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 10 1 10 0.6818464 0.6816239 0.6971852
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 11 2 10 0.7364682 0.6939103 0.7340741
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 12 3 10 0.8017823 0.6693376 0.7238519
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 5 0.6818464 0.6816239 0.6971852
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 5 0.7407227 0.6933761 0.7317037
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 5 0.8029432 0.713141 0.7808889
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6818464 0.6816239 0.6971852
## 2 2 25 0.7466855 0.6762821 0.7271111
## 3 3 25 0.8039853 0.6388889 0.7151111
## 4 1 20 0.6818464 0.6816239 0.6971852
## 5 2 20 0.7338232 0.6976496 0.7420741
## 6 3 20 0.8136485 0.5881410 0.7131852
## 7 1 15 0.6818464 0.6816239 0.6971852
## 8 2 15 0.7436249 0.7035256 0.7508148
## 9 3 15 0.8090775 0.5480769 0.7251852
## 10 1 10 0.6818464 0.6816239 0.6971852
## 11 2 10 0.7364682 0.6939103 0.7340741
## 12 3 10 0.8017823 0.6693376 0.7238519
## 13 1 5 0.6818464 0.6816239 0.6971852
## 14 2 5 0.7407227 0.6933761 0.7317037
## 15 3 5 0.8029432 0.7131410 0.7808889
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 5 0.8029432 0.713141 0.7808889
##
## Observation 1 has a predicted value 0.367
## since this is the weighted average response across the 8 nodes it is a member of:
##
## 1) Node 25, containing 0.5 training observations, with node mean 0.538 and weight 0.327 :
## T2RGH_var <= 350
## T2wSI_predicted = None
## 130 <= ave_T25
##
## 2) Node 29, containing 65 training observations, with node mean 0.0882 and weight 0.247 :
## T2_lesionSIstd <= 91
## 9.5 <= T2min_F_r_i
## ave_T213 <= 220
##
## 3) Node 19, containing 0.5 training observations, with node mean 0.233 and weight 0.189 :
## T2wSI_predicted = None
## T2_lesionSIstd <= 52
## 76 <= ave_T20
##
## 4) Node 36, containing 0.5 training observations, with node mean 0.633 and weight 0.154 :
## T2wSI_predicted = None
## 58 <= ave_T210
## 0.26 <= T2texture_correlation_nondir
##
## 5) Node 27, containing 0.5 training observations, with node mean 0.34 and weight 0.0294 :
## 150 <= T2RGH_var <= 280
## T2wSI_predicted = None
##
## 6) Node 32, containing 92 training observations, with node mean 0.215 and weight 0.0259 :
## T2_lesionSIstd <= 54
## 84 <= ave_T22
##
## 7) Node 44, containing 0.5 training observations, with node mean 0.462 and weight 0.0135 :
## T2wSI_predicted = None
## T2kurt_F_r_i <= 8.8
## T2skew_F_r_i <= 1.3
##
## 8) Node 50, containing 562 training observations, with node mean 0.392 and weight 0.01 :
## ROOT NODE
##
## ============ bagging trees treedata_imgT1
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7734522 0.6784188 0.7070825
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.8239384 0.7136752 0.7553191
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8616809 0.6538462 0.7282133
## maxinter: 5 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 4 5 25 0.8919502 0.6303419 0.7639172
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 1 20 0.7732016 0.6784188 0.7070825
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 2 20 0.8301848 0.6987179 0.7471583
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 7 3 20 0.8600385 0.6917735 0.7803847
## maxinter: 5 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 8 5 20 0.8945424 0.6794872 0.7744098
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 1 15 0.7703389 0.6848291 0.7038764
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 10 2 15 0.828173 0.6800214 0.7483241
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 11 3 15 0.8559951 0.6997863 0.7615855
## maxinter: 5 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 12 5 15 0.898975 0.7019231 0.7904401
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 10 0.7734522 0.6784188 0.7070825
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 10 0.8205744 0.6992521 0.7516759
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 10 0.8673733 0.7163462 0.7754299
## maxinter: 5 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 16 5 10 0.8959078 0.7008547 0.787234
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 17 1 5 0.7734522 0.6784188 0.7070825
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 18 2 5 0.8260425 0.6875 0.7337511
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 19 3 5 0.8511932 0.667735 0.7328767
## maxinter: 5 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 20 5 5 0.8953735 0.6469017 0.7604197
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7734522 0.6784188 0.7070825
## 2 2 25 0.8239384 0.7136752 0.7553191
## 3 3 25 0.8616809 0.6538462 0.7282133
## 4 5 25 0.8919502 0.6303419 0.7639172
## 5 1 20 0.7732016 0.6784188 0.7070825
## 6 2 20 0.8301848 0.6987179 0.7471583
## 7 3 20 0.8600385 0.6917735 0.7803847
## 8 5 20 0.8945424 0.6794872 0.7744098
## 9 1 15 0.7703389 0.6848291 0.7038764
## 10 2 15 0.8281730 0.6800214 0.7483241
## 11 3 15 0.8559951 0.6997863 0.7615855
## 12 5 15 0.8989750 0.7019231 0.7904401
## 13 1 10 0.7734522 0.6784188 0.7070825
## 14 2 10 0.8205744 0.6992521 0.7516759
## 15 3 10 0.8673733 0.7163462 0.7754299
## 16 5 10 0.8959078 0.7008547 0.7872340
## 17 1 5 0.7734522 0.6784188 0.7070825
## 18 2 5 0.8260425 0.6875000 0.7337511
## 19 3 5 0.8511932 0.6677350 0.7328767
## 20 5 5 0.8953735 0.6469017 0.7604197
## maxinter nodesize rocTrain rocValid rocTest
## 12 5 15 0.898975 0.7019231 0.7904401
##
## Observation 1 has a predicted value 0.181
## since this is the weighted average response across the 7 nodes it is a member of:
##
## 1) Node 76, containing 103 training observations, with node mean 0.0485 and weight 0.344 :
## texture_sumvariance_nondir_post2 <= 670
## 350 <= min_F_r_i
## SER_inside <= 0.8
## A_countor <= 3.2
## 0.57 <= dce2SE4
##
## 2) Node 36, containing 23 training observations, with node mean 0.217 and weight 0.252 :
## 480 <= texture_sumvariance_nondir_post2
## dce3SE2 <= 0.78
##
## 3) Node 83, containing 235 training observations, with node mean 0.2 and weight 0.181 :
## texture_sumvariance_nondir_post2 <= 940
## SER_inside <= 0.82
## 0.071 <= max_RGH_var
## -0.044 <= Kpeak_countor
## 0.52 <= max_RGH_mean
##
## 4) Node 41, containing 27 training observations, with node mean 0.481 and weight 0.114 :
## SER_inside <= 0.72
## 210 <= texture_variance_nondir_post3 <= 260
##
## 5) Node 14, containing 12 training observations, with node mean 0.167 and weight 0.0657 :
## SER_inside <= 0.76
## 13 <= V14
## texture_energy_nondir_post3 <= 0.0011
## maxVr_countor <= 0.11
##
## 6) Node 79, containing 139 training observations, with node mean 0.107 and weight 0.033 :
## texture_sumvariance_nondir_post4 <= 1100
## SER_inside <= 0.68
## texture_energy_nondir_post3 <= 0.0087
## 0.67 <= lateSE9
##
## 7) Node 84, containing 564 training observations, with node mean 0.392 and weight 0.01 :
## ROOT NODE
##
## ============ bagging trees treedata_all
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7728652 0.6880342 0.7157756
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.8233645 0.7190171 0.743537
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8564437 0.7104701 0.7908343
## maxinter: 5 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 4 5 25 0.9020092 0.6826923 0.7594007
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 1 20 0.7726211 0.6944444 0.7193008
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 2 20 0.8138002 0.7003205 0.7381022
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 7 3 20 0.8568658 0.7259615 0.7735018
## maxinter: 5 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 8 5 20 0.9076158 0.7275641 0.8134548
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 1 15 0.7731884 0.6928419 0.7137192
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 10 2 15 0.8192948 0.6907051 0.7342832
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 11 3 15 0.856193 0.6997863 0.7714454
## maxinter: 5 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 12 5 15 0.8898658 0.758547 0.8062573
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 10 0.772542 0.6875 0.7156287
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 10 0.8177051 0.6720085 0.7256169
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 10 0.8534821 0.6981838 0.75896
## maxinter: 5 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 16 5 10 0.9015474 0.6987179 0.7925969
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 17 1 5 0.7725882 0.6923077 0.7147474
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 18 2 5 0.8198158 0.7200855 0.7441246
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 19 3 5 0.8519782 0.6570513 0.7439777
## maxinter: 5 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 20 5 5 0.8982164 0.7168803 0.8022914
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7728652 0.6880342 0.7157756
## 2 2 25 0.8233645 0.7190171 0.7435370
## 3 3 25 0.8564437 0.7104701 0.7908343
## 4 5 25 0.9020092 0.6826923 0.7594007
## 5 1 20 0.7726211 0.6944444 0.7193008
## 6 2 20 0.8138002 0.7003205 0.7381022
## 7 3 20 0.8568658 0.7259615 0.7735018
## 8 5 20 0.9076158 0.7275641 0.8134548
## 9 1 15 0.7731884 0.6928419 0.7137192
## 10 2 15 0.8192948 0.6907051 0.7342832
## 11 3 15 0.8561930 0.6997863 0.7714454
## 12 5 15 0.8898658 0.7585470 0.8062573
## 13 1 10 0.7725420 0.6875000 0.7156287
## 14 2 10 0.8177051 0.6720085 0.7256169
## 15 3 10 0.8534821 0.6981838 0.7589600
## 16 5 10 0.9015474 0.6987179 0.7925969
## 17 1 5 0.7725882 0.6923077 0.7147474
## 18 2 5 0.8198158 0.7200855 0.7441246
## 19 3 5 0.8519782 0.6570513 0.7439777
## 20 5 5 0.8982164 0.7168803 0.8022914
## maxinter nodesize rocTrain rocValid rocTest
## 8 5 20 0.9076158 0.7275641 0.8134548
##
## Observation 1 has a predicted value 0.32
## since this is the weighted average response across the 11 nodes it is a member of:
##
## 1) Node 93, containing 45 training observations, with node mean 0.222 and weight 0.175 :
## texture_sumvariance_nondir_post2 <= 600
## 230 <= texture_sumaverage_nondir_post4
## dce3SE4 <= 1.5
## V14 <= 22
##
## 2) Node 83, containing 38 training observations, with node mean 0.289 and weight 0.17 :
## 430 <= texture_sumvariance_nondir_post1
## V13 <= 12
## texture_sumvariance_nondir_post2 <= 1000
## texture_energy_nondir_post4 <= 0.0022
##
## 3) Node 21, containing 14 training observations, with node mean 0.0714 and weight 0.156 :
## Vr_increasingRate_countor <= 0.099
## 430 <= texture_sumvariance_nondir_post1
## lateSE10 <= 1.3
## texture_inversediffmoment_nondir_post1 <= 0.15
##
## 4) Node 105, containing 66 training observations, with node mean 0.788 and weight 0.144 :
## 0.82 <= Slope_ini_countor
## 3.1 <= texture_entropy_nondir_post2
##
## 5) Node 118, containing 95 training observations, with node mean 0.168 and weight 0.119 :
## texture_sumvariance_nondir_post2 <= 620
## Vr_increasingRate_countor <= 0.034
##
## 6) Node 31, containing 17 training observations, with node mean 0.294 and weight 0.0841 :
## 3 <= Slope_ini_countor
## 52 <= ave_T210
##
## 7) Node 78, containing 34 training observations, with node mean 0.647 and weight 0.0613 :
## 420 <= texture_sumvariance_nondir_post1
## V13 <= 14
## 1 <= Slope_ini_countor
##
## 8) Node 125, containing 175 training observations, with node mean 0.137 and weight 0.0423 :
## Vr_increasingRate_countor <= 0.099
## alpha_inside <= 0.49
## A_inside <= 200
##
## 9) Node 66, containing 29 training observations, with node mean 0.207 and weight 0.0208 :
## 400 <= texture_sumvariance_nondir_post1
## V13 <= 12
## dce3SE15 <= 1.5
## Vr_increasingRate_countor <= 0.21
##
## 10) Node 97, containing 54 training observations, with node mean 0.537 and weight 0.0165 :
## alpha_inside <= 0.47
## 3.1 <= texture_entropy_nondir_post2
## 12 <= V14
##
## 11) Node 133, containing 564 training observations, with node mean 0.392 and weight 0.01 :
## ROOT NODE
## id C NC pred obs
## 1 15 0.3110406 0.6889594 NC NC
## 2 16 0.3774373 0.6225627 NC NC
## 3 17 0.4003017 0.5996983 NC NC
## 4 39 0.5324928 0.4675072 C C
## 5 42 0.3621800 0.6378200 NC NC
## 6 43 0.2448005 0.7551995 NC NC
## id C NC pred obs
## 1 15 0.3665715 0.6334285 NC NC
## 2 16 0.4605225 0.5394775 NC NC
## 3 17 0.5551419 0.4448581 C NC
## 4 39 0.5315584 0.4684416 C C
## 5 42 0.1866425 0.8133575 NC NC
## 6 43 0.3609013 0.6390987 NC NC
## id C NC pred obs
## 1 15 0.1811563 0.8188437 NC NC
## 2 16 0.1969295 0.8030705 NC NC
## 3 17 0.4904886 0.5095114 NC NC
## 4 39 0.8685149 0.1314851 C C
## 5 42 0.4985963 0.5014037 NC NC
## 6 43 0.2914756 0.7085244 NC NC
## id C NC pred obs
## 1 15 0.3203306 0.6796694 NC NC
## 2 16 0.2116481 0.7883519 NC NC
## 3 17 0.3197896 0.6802104 NC NC
## 4 39 0.8402422 0.1597578 C C
## 5 42 0.5152374 0.4847626 C NC
## 6 43 0.2386104 0.7613896 NC NC
##
## Call:
## roc.default(response = perf_imgT2$obs, predictor = perf_imgT2$C)
##
## Data: perf_imgT2$C in 323 controls (perf_imgT2$obs C) > 505 cases (perf_imgT2$obs NC).
## Area under the curve: 0.7498
##
## Call:
## roc.default(response = perf_allT2$obs, predictor = perf_allT2$C)
##
## Data: perf_allT2$C in 321 controls (perf_allT2$obs C) > 507 cases (perf_allT2$obs NC).
## Area under the curve: 0.7443
##
## Call:
## roc.default(response = perf_imgT1$obs, predictor = perf_imgT1$C)
##
## Data: perf_imgT1$C in 340 controls (perf_imgT1$obs C) > 488 cases (perf_imgT1$obs NC).
## Area under the curve: 0.8335
##
## Call:
## roc.default(response = perf_all$obs, predictor = perf_all$C)
##
## Data: perf_all$C in 335 controls (perf_all$obs C) > 493 cases (perf_all$obs NC).
## Area under the curve: 0.8519
## Area under the curve: 0.7498
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.3962598 0.5944 0.6471 0.6966 0.701 0.7386 0.7762
## Area under the curve: 0.7443
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.4375099 0.5421 0.595 0.648 0.7495 0.787 0.8205
## Area under the curve: 0.8335
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.4158285 0.6735 0.7206 0.7676 0.7787 0.8156 0.8484
## Area under the curve: 0.8519
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.3770788 0.7343 0.7821 0.8239 0.7485 0.787 0.8235
## massB massM nonmassB nonmassM
## 212 144 128 76
## massB massM nonmassB nonmassM
## 28 24 14 1
## massB massM nonmassB nonmassM
## 212 144 128 76
## massB massM nonmassB nonmassM
## 28 24 14 1
## massB massM nonmassB nonmassM
## 212 144 128 76
## massB massM nonmassB nonmassM
## 28 24 14 1
## 0.03517588 0.05
## Selected features for group: MeanDecreaseGini imgT2
## =========NULL
## [1] "T2RGH_var" "T2texture_entropy_nondir"
## [3] "ave_T211" "T2texture_correlation_nondir"
## [5] "T2texture_inversediffmoment_nondir" "T2grad_margin_var"
## [7] "T2kurt_F_r_i" "ave_T210"
## [9] "T2skew_F_r_i" "T2RGH_mean"
## [11] "T2texture_contrast_nondir" "ave_T21"
## [13] "ave_T218" "T2texture_energy_nondir"
## [15] "ave_T20" "ave_T28"
## [17] "T2texture_sumaverage_nondir" "ave_T216"
## [19] "ave_T219" "ave_T26"
## [21] "T2var_F_r_i" "ave_T29"
## [23] "ave_T214" "T2_lesionSI"
## [25] "T2grad_margin" "T2texture_diffentropy_nondir"
## [27] "ave_T213" "ave_T23"
## [29] "ave_T22" "ave_T217"
## [31] "ave_T24" "ave_T25"
## [33] "T2texture_variance_nondir" "T2min_F_r_i"
## [35] "T2_lesionSIstd"
## -0.05102041 0.05
## Selected features for group: MeanDecreaseGini allT2
## =========NULL
## [1] "T2RGH_var" "ave_T211"
## [3] "T2wSI_predicted" "T2texture_correlation_nondir"
## [5] "T2kurt_F_r_i" "T2skew_F_r_i"
## [7] "T2texture_sumaverage_nondir" "ave_T212"
## [9] "T2grad_margin_var" "ave_T25"
## [11] "LMSIR_predicted" "ave_T217"
## [13] "ave_T28" "T2texture_diffentropy_nondir"
## [15] "T2texture_diffvariance_nondir" "ave_T214"
## [17] "T2var_F_r_i" "ave_T216"
## [19] "ave_T23" "T2_lesionSI"
## [21] "ave_T213" "ave_T29"
## [23] "T2texture_entropy_nondir" "T2min_F_r_i"
## [25] "ave_T218" "T2texture_inversediffmoment_nondir"
## [27] "T2RGH_mean" "T2_lesionSIstd"
## 0.05031447 0.05
## -0.01986755 0.05
## Selected features for group: MeanDecreaseGini imgT1
## =========NULL
## [1] "texture_variance_nondir_post3" "alpha_inside"
## [3] "V18" "max_RGH_mean"
## [5] "iiMin_change_Variance_uptake" "V1"
## [7] "V14" "texture_inversediffmoment_nondir_post2"
## [9] "V2" "texture_sumaverage_nondir_post1"
## [11] "max_RGH_var" "V4"
## [13] "texture_entropy_nondir_post4" "A_inside"
## [15] "iiiMax_Margin_Gradient" "earlySE4"
## [17] "texture_sumentropy_nondir_post1" "dce2SE8"
## [19] "SER_countor" "dce2SE14"
## [21] "dce3SE17" "earlySE7"
## [23] "earlySE3" "dce3SE18"
## [25] "earlySE0" "dce2SE17"
## [27] "min_F_r_i" "max_RGH_mean_k"
## [29] "lateSE1" "dce3SE10"
## [31] "texture_correlation_nondir_post1" "beta_inside"
## 0.04375 0.05
## Selected features for group: MeanDecreaseGini all
## =========NULL
## [1] "texture_sumvariance_nondir_post1" "texture_inversediffmoment_nondir_post3"
## [3] "V8" "texture_diffvariance_nondir_post2"
## [5] "V4" "mean_F_r_i"
## [7] "max_RGH_mean" "V6"
## [9] "earlySE1" "V10"
## [11] "ave_T23" "dce2SE17"
## [13] "texture_correlation_nondir_post4" "T2texture_sumentropy_nondir"
## [15] "T2RGH_mean" "lateSE6"
## [17] "dce2SE11" "maxVr_countor"
## [19] "edge_sharp_mean" "beta_countor"
## [21] "V7" "V5"
## [23] "Slope_ini_countor" "texture_sumaverage_nondir_post1"
## [25] "T2wSI_predicted" "V19"
## [27] "dce3SE13" "iMax_Variance_uptake"
## [29] "T2grad_margin" "V15"
## [31] "earlySE14" "skew_F_r_i"
## [33] "dce3SE4" "T2RGH_var"
## [35] "V1" "Vr_decreasingRate_inside"
## [37] "texture_variance_nondir_post4" "A_inside"
## lesion_id cad_pt_no_txt exam_a_number_txt BIRADS lesion_label
## 6 6 0066 4583735 3 massB
## 7 7 0066 7556910 4 nonmassB
## 13 13 0121 6714524 4 massB
## 14 14 0121 7091267 4 massB
## 18 18 0127 4696964 4 nonmassB
## 20 20 0130 5017534 2 massB
## 21 21 0130 5017534 2 massB
## 22 22 0130 7347205 4 massB
## 38 38 0189 5057674 4 nonmassB
## 67 67 0276 6952525 4 massM
## 68 68 0276 6952525 4 massM
## 84 84 0426 7169326 4 nonmassB
## 85 85 0426 7169326 4 nonmassB
## 86 86 0426 7169326 4 nonmassB
## 111 111 0578 6765702 6 massM
## 126 126 0657 6980780 4 massM
## 129 129 0666 5088826 3 massM
## 136 136 0679 4994641 6 massM
## 211 211 0783 4758418 3 massB
## 218 218 0789 4785741 3 massM
## 222 222 0792 5264066 3 massB
## 223 223 0792 5264066 3 massB
## 224 224 0793 4988020 4 massB
## 225 225 0793 7135216 2 massB
## 234 234 0810 4622489 4 massM
## 240 240 0815 4828432 5 massM
## 241 241 0817 5363917 6 massM
## 271 271 0856 4986174 4 massB
## 272 272 0856 4986174 4 massB
## 273 273 0856 6871177 2 massB
## 277 277 0862 5395314 4 massM
## 280 280 0865 5267535 5 massM
## 281 281 0865 5267535 5 nonmassM
## 308 308 0913 7350757 4 massB
## 315 315 0937 7144673 4 massB
## 331 331 0985 7050619 4 nonmassB
## 335 335 0997 7279207 3 massB
## 400 400 2065 7604632 4 nonmassB
## 409 409 2078 5116776 4 massB
## 421 421 3018 6865137 3 massB
## 427 427 3028 6991592 3 massB
## 434 434 3045 7149704 4 nonmassB
## 442 442 3054 6714946 4 nonmassB
## 443 443 3055 7742700 4 massB
## 444 444 3055 7060620 4 massM
## 445 445 3055 7742700 4 nonmassB
## 446 446 3055 7742700 4 massB
## 447 447 3055 7060620 4 massM
## 448 448 3057 7098623 4 nonmassB
## 449 449 3057 7098623 4 massB
## 485 485 4020 6988975 6 massM
## 497 497 4041 7003893 4 nonmassB
## 553 553 6041 5104414 6 massM
## 554 554 6041 5104414 6 massM
## 562 562 6046 ACC108189 5 massM
## 563 563 6046 ACC108189 5 massM
## 564 564 6046 ACC108189 5 massM
## 565 565 6046 ACC108189 5 massM
## 569 569 6051 5426079 6 massM
## 589 589 6223 7043947 4 massM
## 611 611 7085 7616788 2 massB
## 614 614 7094 7171259 4 nonmassB
## 617 617 7097 6805449 4 massB
## 618 618 7097 6805449 4 massB
## 619 619 7097 7388464 2 massB
## 626 626 7178 7074874 6 massM
## 627 627 7183 7404761 4 massB
## lesion_diagnosis find_t2_signal_int
## 6 BENIGN BREAST TISSUE Hypointense or not seen
## 7 DENSE FIBROSIS None
## 13 ADENOSIS Hypointense or not seen
## 14 BENIGN BREAST TISSUE Hyperintense
## 18 FIBROADENOMA Hyperintense
## 20 ATYPICAL LOBULAR HYPERPLASIA Hyperintense
## 21 ATYPICAL LOBULAR HYPERPLASIA Hyperintense
## 22 ATYPICAL LOBULAR HYPERPLASIA Hypointense or not seen
## 38 SCLEROSING ADENOSIS Hypointense or not seen
## 67 InvasiveDuctal Slightly hyperintense
## 68 InvasiveDuctal Slightly hyperintense
## 84 STROMAL FIBROSIS None
## 85 BENIGN BREAST TISSUE Hypointense or not seen
## 86 STROMAL FIBROSIS None
## 111 InvasiveDuctal None
## 126 InvasiveDuctal Hypointense or not seen
## 129 InsituDuctal Hyperintense
## 136 InvasiveDuctal None
## 211 FIBROCYSTIC Hyperintense
## 218 InsituDuctal Slightly hyperintense
## 222 DUCT PAPILLOMA Slightly hyperintense
## 223 DUCT PAPILLOMA Slightly hyperintense
## 224 FIBROADENOMA Hyperintense
## 225 COMPLEX FIBROEPITHELIAL LESION Hyperintense
## 234 InsituDuctal None
## 240 InvasiveDuctal None
## 241 InvasiveDuctal Hyperintense
## 271 FIBROADENOMA Hyperintense
## 272 FIBROADENOMA Hyperintense
## 273 ATYPICAL DUCTAL HYPERPLASIA Hypointense or not seen
## 277 InsituDuctal None
## 280 InvasiveDuctal None
## 281 InvasiveDuctal None
## 308 ADENOSIS Slightly hyperintense
## 315 FIBROEPITHELIAL Hyperintense
## 331 COLUMNAR CELL CHANGES Hyperintense
## 335 ATYPICAL LOBULAR HYPERPLASIA None
## 400 InsituLobular None
## 409 BENIGN BREAST TISSUE Hyperintense
## 421 FAT NECROSIS Hyperintense
## 427 HYPERPLASIA Hypointense or not seen
## 434 ADENOSIS None
## 442 BENIGN BREAST TISSUE Hypointense or not seen
## 443 COLUMNAR CELL CHANGES Hypointense or not seen
## 444 InvasiveDuctal Hypointense or not seen
## 445 COLUMNAR CELL CHANGES None
## 446 STROMAL HYPERPLASIA Hypointense or not seen
## 447 InvasiveDuctal Hypointense or not seen
## 448 BENIGN BREAST TISSUE None
## 449 TUBULAR ADENOMA Slightly hyperintense
## 485 InvasiveLobular Slightly hyperintense
## 497 BENIGN BREAST TISSUE Slightly hyperintense
## 553 InvasiveDuctal Slightly hyperintense
## 554 InvasiveDuctal None
## 562 InsituDuctal Hypointense or not seen
## 563 InsituDuctal Hypointense or not seen
## 564 InsituDuctal Hyperintense
## 565 InsituDuctal None
## 569 InvasiveDuctal Hyperintense
## 589 InvasiveDuctal Hyperintense
## 611 ATYPICAL DUCTAL HYPERPLASIA None
## 614 BENIGN BREAST TISSUE None
## 617 SCLEROSING ADENOSIS Hypointense or not seen
## 618 FIBROADENOMA None
## 619 ATYPICAL DUCTAL HYPERPLASIA Slightly hyperintense
## 626 InvasiveDuctal Hypointense or not seen
## 627 FIBROCYSTIC Hyperintense
##
## ============ bagging trees treedata_imgT2
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6798596 0.6014286 0.6410675
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.7471858 0.5757143 0.6646242
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8215976 0.5561905 0.6659858
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 4 1 20 0.6798596 0.6014286 0.6410675
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 2 20 0.7628142 0.5657143 0.6666667
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 3 20 0.7970989 0.5452381 0.6647603
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 7 1 15 0.6798596 0.6014286 0.6410675
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 8 2 15 0.7537834 0.3957143 0.6672113
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 3 15 0.8238837 0.4352381 0.6836874
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 10 1 10 0.6798596 0.6014286 0.6410675
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 11 2 10 0.7655882 0.4866667 0.6188725
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 12 3 10 0.7923128 0.5638095 0.6802832
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 5 0.6798596 0.6014286 0.6410675
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 5 0.7603209 0.5466667 0.6456972
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 5 0.8177139 0.5347619 0.6598584
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6798596 0.6014286 0.6410675
## 2 2 25 0.7471858 0.5757143 0.6646242
## 3 3 25 0.8215976 0.5561905 0.6659858
## 4 1 20 0.6798596 0.6014286 0.6410675
## 5 2 20 0.7628142 0.5657143 0.6666667
## 6 3 20 0.7970989 0.5452381 0.6647603
## 7 1 15 0.6798596 0.6014286 0.6410675
## 8 2 15 0.7537834 0.3957143 0.6672113
## 9 3 15 0.8238837 0.4352381 0.6836874
## 10 1 10 0.6798596 0.6014286 0.6410675
## 11 2 10 0.7655882 0.4866667 0.6188725
## 12 3 10 0.7923128 0.5638095 0.6802832
## 13 1 5 0.6798596 0.6014286 0.6410675
## 14 2 5 0.7603209 0.5466667 0.6456972
## 15 3 5 0.8177139 0.5347619 0.6598584
## maxinter nodesize rocTrain rocValid rocTest
## 9 3 15 0.8238837 0.4352381 0.6836874
##
## Observation 1 has a predicted value 0.433
## since this is the weighted average response across the 12 nodes it is a member of:
##
## 1) Node 58, containing 107 training observations, with node mean 0.299 and weight 0.199 :
## T2texture_entropy_nondir <= 3.4
## 42 <= T2texture_sumaverage_nondir
## T2kurt_F_r_i <= 0.28
##
## 2) Node 53, containing 100 training observations, with node mean 0.535 and weight 0.178 :
## 3.2 <= T2texture_entropy_nondir
## 67 <= T2texture_sumaverage_nondir
## 350 <= T2RGH_var
##
## 3) Node 65, containing 129 training observations, with node mean 0.504 and weight 0.143 :
## T2texture_entropy_nondir <= 3.5
## T2texture_variance_nondir <= 360
## 0.26 <= T2texture_correlation_nondir
##
## 4) Node 18, containing 25 training observations, with node mean 0.615 and weight 0.117 :
## 350 <= T2RGH_var
## 56 <= T2grad_margin
## 330 <= ave_T216
##
## 5) Node 62, containing 116 training observations, with node mean 0.195 and weight 0.0852 :
## T2texture_entropy_nondir <= 3.4
## T2texture_inversediffmoment_nondir <= 0.13
## 150 <= ave_T213
##
## 6) Node 74, containing 244 training observations, with node mean 0.441 and weight 0.0776 :
## T2texture_entropy_nondir <= 3.5
## T2texture_contrast_nondir <= 480
## 260 <= T2RGH_var
##
## 7) Node 76, containing 360 training observations, with node mean 0.424 and weight 0.0611 :
## T2min_F_r_i <= 24
## 56 <= ave_T210
## 83 <= ave_T213
##
## 8) Node 66, containing 134 training observations, with node mean 0.388 and weight 0.0589 :
## T2texture_entropy_nondir <= 3.5
## 0.29 <= T2texture_correlation_nondir
##
## 9) Node 72, containing 241 training observations, with node mean 0.432 and weight 0.0378 :
## 260 <= T2RGH_var
## T2texture_entropy_nondir <= 3.4
## T2texture_variance_nondir <= 420
##
## 10) Node 49, containing 93 training observations, with node mean 0.376 and weight 0.019 :
## 350 <= T2RGH_var
## T2RGH_mean <= 50
## 58 <= T2grad_margin
##
## 11) Node 75, containing 284 training observations, with node mean 0.526 and weight 0.0123 :
## 350 <= T2RGH_var
## 46 <= ave_T211
## 56 <= ave_T210
##
## 12) Node 77, containing 556 training observations, with node mean 0.393 and weight 0.01 :
## ROOT NODE
##
## ============ bagging trees treedata_allT2
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6864505 0.3938095 0.635
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.7551136 0.5780952 0.6379167
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.815254 0.6342857 0.7054167
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 4 1 20 0.6864505 0.3938095 0.635
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 2 20 0.7551671 0.5980952 0.6544444
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 3 20 0.8156283 0.5585714 0.6409722
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 7 1 15 0.6864505 0.3938095 0.635
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 8 2 15 0.7712433 0.5585714 0.6516667
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 3 15 0.828623 0.4838095 0.6643056
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 10 1 10 0.6864505 0.3938095 0.635
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 11 2 10 0.7611698 0.6333333 0.6858333
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 12 3 10 0.8084358 0.56 0.6444444
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 5 0.6864505 0.3938095 0.635
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 5 0.756143 0.6095238 0.6541667
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 5 0.8078543 0.5195238 0.62625
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6864505 0.3938095 0.6350000
## 2 2 25 0.7551136 0.5780952 0.6379167
## 3 3 25 0.8152540 0.6342857 0.7054167
## 4 1 20 0.6864505 0.3938095 0.6350000
## 5 2 20 0.7551671 0.5980952 0.6544444
## 6 3 20 0.8156283 0.5585714 0.6409722
## 7 1 15 0.6864505 0.3938095 0.6350000
## 8 2 15 0.7712433 0.5585714 0.6516667
## 9 3 15 0.8286230 0.4838095 0.6643056
## 10 1 10 0.6864505 0.3938095 0.6350000
## 11 2 10 0.7611698 0.6333333 0.6858333
## 12 3 10 0.8084358 0.5600000 0.6444444
## 13 1 5 0.6864505 0.3938095 0.6350000
## 14 2 5 0.7561430 0.6095238 0.6541667
## 15 3 5 0.8078543 0.5195238 0.6262500
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.815254 0.6342857 0.7054167
##
## Observation 1 has a predicted value 0.435
## since this is the weighted average response across the 9 nodes it is a member of:
##
## 1) Node 47, containing 0.5 training observations, with node mean 0.463 and weight 0.304 :
## T2wSI_predicted in {Hyperintense,Hypointense or not seen,Slightly hyperintense}
## 350 <= T2RGH_var
## T2RGH_mean <= 40
##
## 2) Node 32, containing 50 training observations, with node mean 0.5 and weight 0.243 :
## 8700 <= T2var_F_r_i
## T2texture_entropy_nondir <= 3.5
## T2texture_diffvariance_nondir <= 190
##
## 3) Node 48, containing 0.5 training observations, with node mean 0.298 and weight 0.123 :
## T2wSI_predicted in {Hyperintense,Hypointense or not seen,Slightly hyperintense}
## 420 <= T2RGH_var
##
## 4) Node 59, containing 148 training observations, with node mean 0.417 and weight 0.108 :
## 350 <= T2RGH_var
## 46 <= ave_T211
## 170 <= ave_T213
##
## 5) Node 67, containing 0.5 training observations, with node mean 0.258 and weight 0.104 :
## T2texture_entropy_nondir <= 3.4
## T2wSI_predicted in {Hyperintense,Hypointense or not seen,Slightly hyperintense}
## T2texture_diffvariance_nondir <= 340
##
## 6) Node 10, containing 19 training observations, with node mean 0.684 and weight 0.0622 :
## 9 <= T2min_F_r_i
## 220 <= ave_T213
## T2texture_diffvariance_nondir <= 130
##
## 7) Node 65, containing 167 training observations, with node mean 0.382 and weight 0.0265 :
## 22 <= T2RGH_mean
## 49 <= ave_T211
## 190 <= ave_T217
##
## 8) Node 66, containing 0.5 training observations, with node mean 0.366 and weight 0.0194 :
## T2wSI_predicted in {Hyperintense,Hypointense or not seen,Slightly hyperintense}
## 20 <= T2RGH_mean
## 0.1 <= T2texture_correlation_nondir
##
## 9) Node 72, containing 556 training observations, with node mean 0.393 and weight 0.01 :
## ROOT NODE
##
## ============ bagging trees treedata_imgT1
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7619385 0.8142857 0.7713976
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.8321992 0.727619 0.7498646
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8643249 0.627619 0.7195287
## maxinter: 5 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 4 5 25 0.9057286 0.8152381 0.8431744
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 1 20 0.7619385 0.8142857 0.7713976
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 2 20 0.8160829 0.7733333 0.7703142
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 7 3 20 0.8732487 0.7842857 0.8076923
## maxinter: 5 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 8 5 20 0.9024265 0.7419048 0.7990249
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 1 15 0.7614171 0.8133333 0.7711268
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 10 2 15 0.8335227 0.722381 0.7458017
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 11 3 15 0.8629679 0.762381 0.7665222
## maxinter: 5 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 12 5 15 0.8976604 0.587619 0.6966414
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 10 0.7620455 0.8142857 0.7746479
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 10 0.8252741 0.7390476 0.7624594
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 10 0.8617313 0.7904762 0.7948267
## maxinter: 5 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 16 5 10 0.9108356 0.7552381 0.8047129
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 17 1 5 0.7621457 0.8142857 0.7711268
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 18 2 5 0.8159024 0.7633333 0.7730228
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 19 3 5 0.8744051 0.6952381 0.7409263
## maxinter: 5 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 20 5 5 0.904619 0.7352381 0.7811484
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7619385 0.8142857 0.7713976
## 2 2 25 0.8321992 0.7276190 0.7498646
## 3 3 25 0.8643249 0.6276190 0.7195287
## 4 5 25 0.9057286 0.8152381 0.8431744
## 5 1 20 0.7619385 0.8142857 0.7713976
## 6 2 20 0.8160829 0.7733333 0.7703142
## 7 3 20 0.8732487 0.7842857 0.8076923
## 8 5 20 0.9024265 0.7419048 0.7990249
## 9 1 15 0.7614171 0.8133333 0.7711268
## 10 2 15 0.8335227 0.7223810 0.7458017
## 11 3 15 0.8629679 0.7623810 0.7665222
## 12 5 15 0.8976604 0.5876190 0.6966414
## 13 1 10 0.7620455 0.8142857 0.7746479
## 14 2 10 0.8252741 0.7390476 0.7624594
## 15 3 10 0.8617313 0.7904762 0.7948267
## 16 5 10 0.9108356 0.7552381 0.8047129
## 17 1 5 0.7621457 0.8142857 0.7711268
## 18 2 5 0.8159024 0.7633333 0.7730228
## 19 3 5 0.8744051 0.6952381 0.7409263
## 20 5 5 0.9046190 0.7352381 0.7811484
## maxinter nodesize rocTrain rocValid rocTest
## 4 5 25 0.9057286 0.8152381 0.8431744
##
## Observation 1 has a predicted value 0.394
## since this is the weighted average response across the 11 nodes it is a member of:
##
## 1) Node 21, containing 14 training observations, with node mean 0.714 and weight 0.373 :
## 200 <= texture_variance_nondir_post3 <= 280
## alpha_inside <= 0.0038
## 330 <= min_F_r_i
##
## 2) Node 131, containing 208 training observations, with node mean 0.173 and weight 0.179 :
## V18 <= 18
## texture_sumentropy_nondir_post1 <= 2
## SER_countor <= 0.84
## V4 <= 30
## beta_inside <= 0.11
##
## 3) Node 129, containing 164 training observations, with node mean 0.139 and weight 0.123 :
## earlySE7 <= 1.4
## 350 <= min_F_r_i
## alpha_inside <= 0.49
## 2.6 <= V4
## texture_sumaverage_nondir_post1 <= 220
##
## 4) Node 133, containing 262 training observations, with node mean 0.205 and weight 0.113 :
## texture_variance_nondir_post3 <= 340
## alpha_inside <= 0.57
## dce2SE8 <= 2
## 270 <= min_F_r_i
##
## 5) Node 123, containing 125 training observations, with node mean 0.176 and weight 0.0447 :
## SER_countor <= 0.7
## 1.5 <= texture_sumentropy_nondir_post1
## V2 <= 28
## V18 <= 19
## texture_variance_nondir_post3 <= 400
##
## 6) Node 24, containing 15 training observations, with node mean 0.0667 and weight 0.04 :
## 390 <= min_F_r_i
## earlySE4 <= 1.3
## 250 <= texture_variance_nondir_post3
## dce3SE18 <= 0.98
##
## 7) Node 96, containing 45 training observations, with node mean 0.378 and weight 0.037 :
## 260 <= min_F_r_i
## texture_sumentropy_nondir_post1 <= 1.9
## max_RGH_mean <= 0.54
## 0.089 <= iiiMax_Margin_Gradient
## texture_inversediffmoment_nondir_post2 <= 0.18
##
## 8) Node 39, containing 18 training observations, with node mean 0.444 and weight 0.0356 :
## SER_countor <= 0.71
## max_RGH_mean <= 0.53
## V18 <= 31
## 170 <= texture_sumaverage_nondir_post1
## 12 <= V2
##
## 9) Node 46, containing 21 training observations, with node mean 0.286 and weight 0.0296 :
## SER_countor <= 0.71
## 1.7 <= texture_sumentropy_nondir_post1
## V18 <= 30
## max_RGH_mean <= 0.57
## 3.4 <= texture_entropy_nondir_post4
##
## 10) Node 130, containing 197 training observations, with node mean 0.198 and weight 0.012 :
## 0.2 <= iiMin_change_Variance_uptake
## alpha_inside <= 0.56
## texture_sumentropy_nondir_post1 <= 1.8
##
## 11) Node 134, containing 560 training observations, with node mean 0.393 and weight 0.01 :
## ROOT NODE
##
## ============ bagging trees treedata_all
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7463703 0.7033333 0.7820964
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.8102273 0.627619 0.7601571
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8685829 0.7133333 0.8225894
## maxinter: 5 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 4 5 25 0.9112634 0.6990476 0.8328819
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 1 20 0.7463703 0.7033333 0.7820964
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 2 20 0.8122794 0.6609524 0.8006501
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 7 3 20 0.8702473 0.747619 0.8361322
## maxinter: 5 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 8 5 20 0.9038904 0.6557143 0.7864301
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 1 15 0.7463703 0.7033333 0.7820964
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 10 2 15 0.8210963 0.6247619 0.7642199
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 11 3 15 0.8579144 0.6809524 0.8068797
## maxinter: 5 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 12 5 15 0.9064104 0.687619 0.8066089
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 10 0.7462233 0.702381 0.781013
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 10 0.8187032 0.6614286 0.7778982
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 10 0.861143 0.7209524 0.8080986
## maxinter: 5 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 16 5 10 0.912246 0.6914286 0.8228602
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 17 1 5 0.7463703 0.7033333 0.7820964
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 18 2 5 0.8186297 0.7147619 0.8174431
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 19 3 5 0.8696858 0.677619 0.8045775
## maxinter: 5 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 20 5 5 0.9030615 0.7009524 0.8293608
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7463703 0.7033333 0.7820964
## 2 2 25 0.8102273 0.6276190 0.7601571
## 3 3 25 0.8685829 0.7133333 0.8225894
## 4 5 25 0.9112634 0.6990476 0.8328819
## 5 1 20 0.7463703 0.7033333 0.7820964
## 6 2 20 0.8122794 0.6609524 0.8006501
## 7 3 20 0.8702473 0.7476190 0.8361322
## 8 5 20 0.9038904 0.6557143 0.7864301
## 9 1 15 0.7463703 0.7033333 0.7820964
## 10 2 15 0.8210963 0.6247619 0.7642199
## 11 3 15 0.8579144 0.6809524 0.8068797
## 12 5 15 0.9064104 0.6876190 0.8066089
## 13 1 10 0.7462233 0.7023810 0.7810130
## 14 2 10 0.8187032 0.6614286 0.7778982
## 15 3 10 0.8611430 0.7209524 0.8080986
## 16 5 10 0.9122460 0.6914286 0.8228602
## 17 1 5 0.7463703 0.7033333 0.7820964
## 18 2 5 0.8186297 0.7147619 0.8174431
## 19 3 5 0.8696858 0.6776190 0.8045775
## 20 5 5 0.9030615 0.7009524 0.8293608
## maxinter nodesize rocTrain rocValid rocTest
## 7 3 20 0.8702473 0.747619 0.8361322
##
## Observation 1 has a predicted value 0.246
## since this is the weighted average response across the 14 nodes it is a member of:
##
## 1) Node 77, containing 0.5 training observations, with node mean 0.219 and weight 0.156 :
## T2wSI_predicted in {Hyperintense,Hypointense or not seen,Slightly hyperintense}
## dce3SE13 <= 1.6
##
## 2) Node 81, containing 278 training observations, with node mean 0.198 and weight 0.138 :
## texture_sumvariance_nondir_post1 <= 380
## texture_inversediffmoment_nondir_post3 <= 0.24
## V15 <= 41
##
## 3) Node 66, containing 140 training observations, with node mean 0.179 and weight 0.118 :
## earlySE14 <= 0.81
## 1.6 <= A_inside
##
## 4) Node 78, containing 242 training observations, with node mean 0.219 and weight 0.0918 :
## earlySE1 <= 0.87
## texture_sumvariance_nondir_post1 <= 430
##
## 5) Node 65, containing 0.5 training observations, with node mean 0.217 and weight 0.0794 :
## T2wSI_predicted in {Hyperintense,Hypointense or not seen,Slightly hyperintense}
## 22 <= T2RGH_mean
## texture_sumvariance_nondir_post1 <= 400
##
## 6) Node 63, containing 127 training observations, with node mean 0.433 and weight 0.0752 :
## texture_sumvariance_nondir_post1 <= 520
## max_RGH_mean <= 0.54
##
## 7) Node 73, containing 195 training observations, with node mean 0.2 and weight 0.0712 :
## earlySE1 <= 0.87
## texture_sumvariance_nondir_post1 <= 430
## V8 <= 28
##
## 8) Node 80, containing 260 training observations, with node mean 0.375 and weight 0.0679 :
## earlySE14 <= 1.3
## 26 <= T2RGH_mean
## skew_F_r_i <= 0.67
##
## 9) Node 79, containing 253 training observations, with node mean 0.221 and weight 0.0601 :
## Slope_ini_countor <= 0.86
## V1 <= 20
## V8 <= 31
##
## 10) Node 40, containing 61 training observations, with node mean 0.213 and weight 0.0596 :
## 200 <= texture_variance_nondir_post4
## texture_sumvariance_nondir_post1 <= 490
## beta_countor <= 0.086
##
## 11) Node 38, containing 53 training observations, with node mean 0.453 and weight 0.0319 :
## V8 <= 18
## Slope_ini_countor <= 0.78
## max_RGH_mean <= 0.54
##
## 12) Node 76, containing 214 training observations, with node mean 0.201 and weight 0.0297 :
## texture_diffvariance_nondir_post2 <= 150
## texture_sumvariance_nondir_post1 <= 760
## texture_correlation_nondir_post4 <= 0.36
##
## 13) Node 74, containing 203 training observations, with node mean 0.32 and weight 0.0117 :
## 160 <= texture_sumvariance_nondir_post1 <= 430
## texture_inversediffmoment_nondir_post3 <= 0.28
##
## 14) Node 82, containing 560 training observations, with node mean 0.393 and weight 0.01 :
## ROOT NODE
## id C NC pred obs
## 1 6 0.4327367 0.5672633 NC NC
## 2 7 0.6174554 0.3825446 C NC
## 3 13 0.2935430 0.7064570 NC NC
## 4 14 0.5062728 0.4937272 C NC
## 5 18 0.3687135 0.6312865 NC NC
## 6 20 0.3713423 0.6286577 NC NC
## id C NC pred obs
## 1 6 0.4345496 0.5654504 NC NC
## 2 7 0.5525158 0.4474842 C NC
## 3 13 0.2821364 0.7178636 NC NC
## 4 14 0.3770787 0.6229213 NC NC
## 5 18 0.2536531 0.7463469 NC NC
## 6 20 0.2335381 0.7664619 NC NC
## id C NC pred obs
## 1 6 0.3940347 0.6059653 NC NC
## 2 7 0.1922980 0.8077020 NC NC
## 3 13 0.2482597 0.7517403 NC NC
## 4 14 0.3553344 0.6446656 NC NC
## 5 18 0.3154823 0.6845177 NC NC
## 6 20 0.1514887 0.8485113 NC NC
## id C NC pred obs
## 1 6 0.2461934 0.7538066 NC NC
## 2 7 0.3086700 0.6913300 NC NC
## 3 13 0.3512780 0.6487220 NC NC
## 4 14 0.2732974 0.7267026 NC NC
## 5 18 0.2141097 0.7858903 NC NC
## 6 20 0.2145896 0.7854104 NC NC
##
## Call:
## roc.default(response = perf_imgT2$obs, predictor = perf_imgT2$C)
##
## Data: perf_imgT2$C in 374 controls (perf_imgT2$obs C) > 577 cases (perf_imgT2$obs NC).
## Area under the curve: 0.7404
##
## Call:
## roc.default(response = perf_allT2$obs, predictor = perf_allT2$C)
##
## Data: perf_allT2$C in 369 controls (perf_allT2$obs C) > 582 cases (perf_allT2$obs NC).
## Area under the curve: 0.7388
##
## Call:
## roc.default(response = perf_imgT1$obs, predictor = perf_imgT1$C)
##
## Data: perf_imgT1$C in 392 controls (perf_imgT1$obs C) > 559 cases (perf_imgT1$obs NC).
## Area under the curve: 0.8339
##
## Call:
## roc.default(response = perf_all$obs, predictor = perf_all$C)
##
## Data: perf_all$C in 387 controls (perf_all$obs C) > 564 cases (perf_all$obs NC).
## Area under the curve: 0.8485
## Area under the curve: 0.7404
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.3959614 0.5882 0.6364 0.6872 0.6984 0.7331 0.7712
## Area under the curve: 0.7388
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.4322865 0.5528 0.6043 0.6558 0.7234 0.7595 0.7921
## Area under the curve: 0.8339
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.4158285 0.6684 0.7117 0.7577 0.7889 0.8211 0.8533
## Area under the curve: 0.8485
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.409165 0.6615 0.708 0.7545 0.8262 0.8564 0.883
## massB massM nonmassB nonmassM
## 209 155 125 70
## massB massM nonmassB nonmassM
## 31 13 17 7
## massB massM nonmassB nonmassM
## 209 155 125 70
## massB massM nonmassB nonmassM
## 31 13 17 7
## massB massM nonmassB nonmassM
## 209 155 125 70
## massB massM nonmassB nonmassM
## 31 13 17 7
## -0.03196347 0.05
## Selected features for group: MeanDecreaseGini imgT2
## =========NULL
## [1] "T2RGH_var" "T2texture_correlation_nondir"
## [3] "T2kurt_F_r_i" "T2texture_entropy_nondir"
## [5] "T2grad_margin_var" "T2skew_F_r_i"
## [7] "ave_T20" "ave_T25"
## [9] "ave_T23" "T2texture_contrast_nondir"
## [11] "ave_T26" "ave_T214"
## [13] "T2_lesionSIstd" "ave_T212"
## [15] "ave_T216" "T2texture_sumvariance_nondir"
## [17] "ave_T24" "T2min_F_r_i"
## [19] "T2texture_inversediffmoment_nondir" "ave_T211"
## [21] "ave_T28" "ave_T213"
## [23] "T2_lesionSI" "ave_T215"
## [25] "T2texture_sumaverage_nondir" "ave_T210"
## [27] "T2max_F_r_i" "ave_T21"
## [29] "T2texture_sumentropy_nondir" "ave_T218"
## [31] "T2RGH_mean" "T2mean_F_r_i"
## 0.06912442 0.05
## -0.02970297 0.05
## Selected features for group: MeanDecreaseGini allT2
## =========NULL
## [1] "T2RGH_mean" "T2kurt_F_r_i"
## [3] "T2RGH_var" "T2texture_correlation_nondir"
## [5] "ave_T211" "T2texture_energy_nondir"
## [7] "T2_lesionSIstd" "ave_T21"
## [9] "T2max_F_r_i" "T2grad_margin_var"
## [11] "T2wSI_predicted" "T2skew_F_r_i"
## [13] "ave_T20" "LMSIR_predicted"
## [15] "ave_T23" "T2texture_sumentropy_nondir"
## [17] "ave_T22" "ave_T216"
## [19] "ave_T219" "T2grad_margin"
## [21] "ave_T210" "ave_T25"
## [23] "ave_T218" "ave_T214"
## [25] "ave_T24" "T2min_F_r_i"
## [27] "T2texture_inversediffmoment_nondir" "T2texture_sumaverage_nondir"
## [29] "ave_T212" "ave_T27"
## [31] "ave_T217" "ave_T29"
## [33] "ave_T26" "ave_T28"
## [35] "T2texture_sumvariance_nondir" "T2_lesionSI"
## 0.06832298 0.05
## -0.006666667 0.05
## Selected features for group: MeanDecreaseGini imgT1
## =========NULL
## [1] "texture_sumvariance_nondir_post1" "V14"
## [3] "Tpeak_inside" "V18"
## [5] "texture_inversediffmoment_nondir_post3" "texture_diffentropy_nondir_post3"
## [7] "alpha_countor" "skew_F_r_i"
## [9] "texture_sumentropy_nondir_post3" "texture_sumaverage_nondir_post1"
## [11] "ivVariance" "beta_inside"
## [13] "maxVr_countor" "V17"
## [15] "UptakeRate_inside" "texture_correlation_nondir_post3"
## [17] "lateSE18" "dce3SE12"
## [19] "V3" "earlySE0"
## [21] "lateSE5" "dce3SE4"
## [23] "irregularity" "Vr_decreasingRate_countor"
## [25] "dce3SE17" "edge_sharp_mean"
## [27] "earlySE14" "texture_contrast_nondir_post4"
## [29] "k_Max_Margin_Grad" "dce3SE3"
## [31] "texture_sumentropy_nondir_post2" "A_inside"
## 0.0375 0.05
## Selected features for group: MeanDecreaseGini all
## =========NULL
## [1] "irregularity" "UptakeRate_inside"
## [3] "texture_inversediffmoment_nondir_post4" "max_F_r_i"
## [5] "dce2SE19" "Kpeak_inside"
## [7] "iiiMax_Margin_Gradient" "V1"
## [9] "dce2SE10" "ave_T210"
## [11] "texture_diffentropy_nondir_post4" "T2texture_inversediffmoment_nondir"
## [13] "lateSE12" "ave_T20"
## [15] "maxVr_countor" "earlySE1"
## [17] "min_F_r_i" "lateSE7"
## [19] "ave_T21" "lateSE8"
## [21] "earlySE17" "texture_diffentropy_nondir_post1"
## [23] "T2_lesionSIstd" "lateSE0"
## [25] "V6" "ave_T24"
## [27] "V8" "T2grad_margin"
## [29] "T2min_F_r_i" "V19"
## [31] "lateSE1" "V3"
## [33] "texture_diffvariance_nondir_post2" "V14"
## [35] "dce3SE15" "T2mean_F_r_i"
## [37] "A_inside"
## lesion_id cad_pt_no_txt exam_a_number_txt BIRADS lesion_label
## 4 4 0027 7171944 4 nonmassB
## 5 5 0027 6805483 4 nonmassM
## 8 8 0093 7156466 4 nonmassM
## 9 9 0093 7156466 4 nonmassM
## 26 26 0135 7777131 4 massB
## 27 27 0135 5083620 4 nonmassB
## 30 30 0171 4751079 4 massM
## 31 31 0172 4703102 4 massB
## 48 48 0198 4809893 2 massB
## 49 49 0198 4809893 2 massB
## 60 60 0252 5142106 4 massB
## 61 61 0252 5142106 4 massB
## 62 62 0252 6700964 3 nonmassB
## 63 63 0252 6700964 3 massB
## 65 65 0259 7364573 2 nonmassB
## 98 98 0513 5043867 4 nonmassB
## 106 106 0571 4902166 4 massM
## 107 107 0571 4902166 4 nonmassB
## 120 120 0616 7910718 2 massM
## 157 157 0710 5282770 4 massB
## 158 158 0710 5282770 5 massB
## 159 159 0710 6798490 2 massB
## 170 170 0723 4884108 6 massM
## 181 181 0734 4532660 4 massB
## 248 248 0830 4863868 5 massB
## 249 249 0830 4863868 5 massB
## 261 261 0850 5380609 5 massB
## 262 262 0850 5380609 5 nonmassB
## 263 263 0850 5380609 5 massB
## 266 266 0853 4798586 2 nonmassB
## 267 267 0853 4745782 3 nonmassB
## 268 268 0853 6696534 4 nonmassM
## 269 269 0855 4641315 6 massB
## 270 270 0855 4641315 6 nonmassB
## 278 278 0863 4969136 4 massB
## 279 279 0863 4969136 4 massM
## 298 298 0883 5177385 5 nonmassM
## 306 306 0900 6699226 4 massB
## 307 307 0904 7133915 3 massB
## 349 349 1027 6930730 3 nonmassB
## 351 351 1045 7231265 4 massB
## 353 353 1026 6907382 4 massB
## 354 354 1053 7748055 4 massB
## 358 358 1065 7741665 4 nonmassB
## 375 375 1099 7646705 5 massM
## 376 376 1099 7646705 4 nonmassM
## 402 402 2069 4976319 6 massM
## 426 426 3026 6830523 4 massB
## 431 431 3035 7002031 4 massB
## 432 432 3035 7145247 4 massB
## 438 438 3052 7100200 4 massB
## 439 439 3052 7100200 4 massB
## 459 459 3076 7053450 6 massM
## 460 460 3076 7053450 6 massM
## 495 495 4040 7003416 6 massM
## 496 496 4040 7085105 4 massB
## 504 504 4049 7009602 6 massM
## 519 519 6019 ACC109175 4 massM
## 537 537 6032 4982490 4 nonmassB
## 546 546 6038 5044471 6 massM
## 547 547 6038 5044471 6 nonmassB
## 548 548 6038 5044471 6 nonmassB
## 581 581 6105 5069712 4 nonmassM
## 585 585 6148 7446343 4 massB
## 590 590 6224 4559525 4 nonmassB
## 591 591 6226 6718391 4 massB
## 603 603 7053 7956343 4 nonmassB
## 622 622 7127 6989740 4 massB
## lesion_diagnosis find_t2_signal_int
## 4 STROMAL FIBROSIS None
## 5 InsituDuctal None
## 8 InvasiveDuctal None
## 9 InvasiveDuctal None
## 26 FIBROCYSTIC None
## 27 FIBROCYSTIC Hypointense or not seen
## 30 InsituDuctal Hypointense or not seen
## 31 FIBROCYSTIC None
## 48 FIBROCYSTIC None
## 49 FIBROADENOMA None
## 60 FIBROADENOMA Hypointense or not seen
## 61 FIBROADENOMA Hypointense or not seen
## 62 BENIGN BREAST TISSUE Hypointense or not seen
## 63 BENIGN BREAST TISSUE Hyperintense
## 65 BENIGN BREAST TISSUE None
## 98 ATYPICAL DUCTAL HYPERPLASIA None
## 106 InvasiveDuctal Hypointense or not seen
## 107 DUCT PAPILLOMA Hypointense or not seen
## 120 InsituDuctal Hypointense or not seen
## 157 FIBROADENOMA Hyperintense
## 158 DUCT PAPILLOMA Hyperintense
## 159 DUCTAL HYPERPLASIA None
## 170 InsituDuctal None
## 181 ATYPICAL DUCTAL HYPERPLASIA Slightly hyperintense
## 248 ATYPICAL DUCTAL HYPERPLASIA None
## 249 Cyst None
## 261 BENIGN BREAST TISSUE Slightly hyperintense
## 262 ATYPICAL DUCTAL HYPERPLASIA None
## 263 FIBROCYSTIC Slightly hyperintense
## 266 FIBROCYSTIC None
## 267 FIBROCYSTIC Slightly hyperintense
## 268 InsituDuctal Slightly hyperintense
## 269 FIBROSIS Hyperintense
## 270 FIBROSIS None
## 278 DUCT PAPILLOMA Hyperintense
## 279 InvasiveLobular Hypointense or not seen
## 298 InsituDuctal None
## 306 Cyst None
## 307 FIBROCYSTIC Hyperintense
## 349 FIBROADENOMA None
## 351 BENIGN BREAST TISSUE Hypointense or not seen
## 353 DENSE FIBROSIS AND FIBROADENOMATOID CHANGE Hyperintense
## 354 INFLAMED CYST WALL Hyperintense
## 358 BENIGN BREAST TISSUE None
## 375 InsituDuctal Hypointense or not seen
## 376 InsituDuctal None
## 402 InsituDuctal Hypointense or not seen
## 426 FIBROCYSTIC Hypointense or not seen
## 431 FIBROADENOMA Hyperintense
## 432 FIBROCYSTIC None
## 438 STROMAL FIBROSIS Hyperintense
## 439 FIBROADENOMA Hyperintense
## 459 InsituDuctal Hypointense or not seen
## 460 InsituDuctal Hypointense or not seen
## 495 InvasiveDuctal None
## 496 FIBROADENOMA Hyperintense
## 504 InvasiveDuctal None
## 519 InvasiveLobular Hypointense or not seen
## 537 FIBROCYSTIC None
## 546 InvasiveDuctal None
## 547 BENIGN BREAST TISSUE None
## 548 BENIGN BREAST TISSUE Hyperintense
## 581 InsituDuctal None
## 585 SCLEROSING ADENOSIS Hypointense or not seen
## 590 FIBROCYSTIC Hyperintense
## 591 BENIGN BREAST TISSUE Slightly hyperintense
## 603 FIBROTIC STROMA Hypointense or not seen
## 622 BENIGN BREAST TISSUE Hypointense or not seen
##
## ============ bagging trees treedata_imgT2
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6953227 0.5291667 0.5818815
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.7608184 0.5729167 0.6248548
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8098736 0.5739583 0.6880081
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 4 1 20 0.6953227 0.5291667 0.5818815
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 2 20 0.7606188 0.5380208 0.6033682
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 3 20 0.8150898 0.559375 0.6811847
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 7 1 15 0.6953227 0.5317708 0.5827526
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 8 2 15 0.7538789 0.5265625 0.6132404
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 3 15 0.8123886 0.5447917 0.642712
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 10 1 10 0.6953227 0.5317708 0.5827526
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 11 2 10 0.7606853 0.5354167 0.6103368
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 12 3 10 0.8043713 0.5109375 0.6370499
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 5 0.6953227 0.5317708 0.5827526
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 5 0.7597139 0.525 0.620935
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 5 0.8097605 0.55 0.6855401
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6953227 0.5291667 0.5818815
## 2 2 25 0.7608184 0.5729167 0.6248548
## 3 3 25 0.8098736 0.5739583 0.6880081
## 4 1 20 0.6953227 0.5291667 0.5818815
## 5 2 20 0.7606188 0.5380208 0.6033682
## 6 3 20 0.8150898 0.5593750 0.6811847
## 7 1 15 0.6953227 0.5317708 0.5827526
## 8 2 15 0.7538789 0.5265625 0.6132404
## 9 3 15 0.8123886 0.5447917 0.6427120
## 10 1 10 0.6953227 0.5317708 0.5827526
## 11 2 10 0.7606853 0.5354167 0.6103368
## 12 3 10 0.8043713 0.5109375 0.6370499
## 13 1 5 0.6953227 0.5317708 0.5827526
## 14 2 5 0.7597139 0.5250000 0.6209350
## 15 3 5 0.8097605 0.5500000 0.6855401
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8098736 0.5739583 0.6880081
##
## Observation 1 has a predicted value 0.404
## since this is the weighted average response across the 11 nodes it is a member of:
##
## 1) Node 73, containing 252 training observations, with node mean 0.408 and weight 0.228 :
## 22 <= T2RGH_mean
## T2texture_correlation_nondir <= 0.27
## 50 <= ave_T216
##
## 2) Node 3, containing 13 training observations, with node mean 0.0769 and weight 0.168 :
## 3.4 <= T2texture_entropy_nondir
## T2texture_inversediffmoment_nondir <= 0.068
## T2texture_sumvariance_nondir <= 1000
##
## 3) Node 40, containing 74 training observations, with node mean 0.28 and weight 0.144 :
## 2.5 <= T2min_F_r_i
## 260 <= T2RGH_var
## ave_T20 <= 200
##
## 4) Node 71, containing 239 training observations, with node mean 0.427 and weight 0.121 :
## T2min_F_r_i <= 6.5
## T2texture_entropy_nondir <= 3.5
## ave_T23 <= 200
##
## 5) Node 55, containing 109 training observations, with node mean 0.634 and weight 0.119 :
## 54 <= ave_T210
## 350 <= T2RGH_var
## ave_T211 <= 150
##
## 6) Node 26, containing 36 training observations, with node mean 0.389 and weight 0.0545 :
## 3.5 <= T2texture_entropy_nondir
## T2texture_inversediffmoment_nondir <= 0.07
## ave_T215 <= 280
##
## 7) Node 28, containing 37 training observations, with node mean 0.757 and weight 0.0516 :
## 3.4 <= T2texture_entropy_nondir
## 1.5 <= T2kurt_F_r_i
##
## 8) Node 18, containing 25 training observations, with node mean 0.885 and weight 0.0403 :
## 350 <= T2RGH_var
## 3.5 <= T2texture_entropy_nondir
## 1.5 <= T2kurt_F_r_i
##
## 9) Node 51, containing 99 training observations, with node mean 0.426 and weight 0.0363 :
## 54 <= ave_T210
## 350 <= T2RGH_var
## 4800 <= T2grad_margin_var
##
## 10) Node 70, containing 201 training observations, with node mean 0.581 and weight 0.0264 :
## 350 <= T2RGH_var
## 380 <= T2max_F_r_i
## ave_T23 <= 320
##
## 11) Node 75, containing 555 training observations, with node mean 0.403 and weight 0.01 :
## ROOT NODE
##
## ============ bagging trees treedata_allT2
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7015968 0.4744792 0.6013494
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.7597206 0.5229167 0.6356589
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8167132 0.5182292 0.6643698
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 4 1 20 0.7015968 0.4744792 0.6013494
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 2 20 0.7568197 0.5036458 0.6310652
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 3 20 0.81169 0.5291667 0.6567614
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 7 1 15 0.7015968 0.4744792 0.6013494
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 8 2 15 0.761477 0.5276042 0.6243181
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 3 15 0.8246707 0.5385417 0.655613
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 10 1 10 0.7015968 0.4744792 0.6013494
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 11 2 10 0.7600798 0.4890625 0.6323572
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 12 3 10 0.8171524 0.5260417 0.6537468
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 5 0.7015968 0.4744792 0.6013494
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 5 0.7576248 0.4973958 0.654321
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 5 0.8179441 0.49375 0.6837496
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7015968 0.4744792 0.6013494
## 2 2 25 0.7597206 0.5229167 0.6356589
## 3 3 25 0.8167132 0.5182292 0.6643698
## 4 1 20 0.7015968 0.4744792 0.6013494
## 5 2 20 0.7568197 0.5036458 0.6310652
## 6 3 20 0.8116900 0.5291667 0.6567614
## 7 1 15 0.7015968 0.4744792 0.6013494
## 8 2 15 0.7614770 0.5276042 0.6243181
## 9 3 15 0.8246707 0.5385417 0.6556130
## 10 1 10 0.7015968 0.4744792 0.6013494
## 11 2 10 0.7600798 0.4890625 0.6323572
## 12 3 10 0.8171524 0.5260417 0.6537468
## 13 1 5 0.7015968 0.4744792 0.6013494
## 14 2 5 0.7576248 0.4973958 0.6543210
## 15 3 5 0.8179441 0.4937500 0.6837496
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 5 0.8179441 0.49375 0.6837496
##
## Observation 1 has a predicted value 0.533
## since this is the weighted average response across the 10 nodes it is a member of:
##
## 1) Node 21, containing 0.5 training observations, with node mean 0.379 and weight 0.384 :
## T2wSI_predicted = None
## 390 <= T2max_F_r_i
## 4.6 <= T2kurt_F_r_i
##
## 2) Node 41, containing 81 training observations, with node mean 0.699 and weight 0.186 :
## 350 <= T2RGH_var
## 50 <= ave_T211 <= 130
##
## 3) Node 44, containing 0.5 training observations, with node mean 0.682 and weight 0.16 :
## 410 <= T2max_F_r_i
## T2wSI_predicted = None
## ave_T27 <= 220
##
## 4) Node 59, containing 249 training observations, with node mean 0.558 and weight 0.0866 :
## 350 <= T2RGH_var
## T2RGH_mean <= 50
## 56 <= ave_T210
##
## 5) Node 58, containing 231 training observations, with node mean 0.566 and weight 0.075 :
## 350 <= T2RGH_var
## 54 <= ave_T210
## ave_T28 <= 280
##
## 6) Node 38, containing 74 training observations, with node mean 0.547 and weight 0.0465 :
## 260 <= T2RGH_var
## T2texture_correlation_nondir <= 0.26
## T2texture_energy_nondir <= 0.00063
##
## 7) Node 54, containing 0.5 training observations, with node mean 0.626 and weight 0.0316 :
## T2wSI_predicted = None
## 390 <= T2max_F_r_i
## 230 <= T2RGH_var
##
## 8) Node 60, containing 356 training observations, with node mean 0.511 and weight 0.0114 :
## 22 <= T2RGH_mean
## 56 <= ave_T210
## ave_T23 <= 320
##
## 9) Node 61, containing 555 training observations, with node mean 0.403 and weight 0.01 :
## ROOT NODE
##
## 10) Node 22, containing 29 training observations, with node mean 0.241 and weight 0.00873 :
## T2texture_energy_nondir <= 0.00054
## 140 <= ave_T219
## ave_T210 <= 180
##
## ============ bagging trees treedata_imgT1
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7766267 0.7609375 0.7842809
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.8338656 0.7171875 0.7805184
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8670659 0.7177083 0.7911093
## maxinter: 5 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 4 5 25 0.9010778 0.6354167 0.7828874
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 1 20 0.7757552 0.771875 0.7897157
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 2 20 0.8327878 0.7546875 0.7909699
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 7 3 20 0.8759681 0.6208333 0.7447046
## maxinter: 5 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 8 5 20 0.904145 0.7041667 0.8010033
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 1 15 0.7766267 0.7609375 0.7842809
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 10 2 15 0.8404059 0.7333333 0.7929208
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 11 3 15 0.8709914 0.7375 0.7899944
## maxinter: 5 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 12 5 15 0.9007186 0.7083333 0.8124303
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 10 0.7766267 0.7609375 0.7842809
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 10 0.834644 0.7098958 0.7689521
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 10 0.8657152 0.7635417 0.8131271
## maxinter: 5 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 16 5 10 0.9039188 0.6161458 0.7660256
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 17 1 5 0.7766267 0.7609375 0.7842809
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 18 2 5 0.8414039 0.7442708 0.7867893
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 19 3 5 0.8710512 0.7041667 0.7806577
## maxinter: 5 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 20 5 5 0.9132535 0.7125 0.8113155
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7766267 0.7609375 0.7842809
## 2 2 25 0.8338656 0.7171875 0.7805184
## 3 3 25 0.8670659 0.7177083 0.7911093
## 4 5 25 0.9010778 0.6354167 0.7828874
## 5 1 20 0.7757552 0.7718750 0.7897157
## 6 2 20 0.8327878 0.7546875 0.7909699
## 7 3 20 0.8759681 0.6208333 0.7447046
## 8 5 20 0.9041450 0.7041667 0.8010033
## 9 1 15 0.7766267 0.7609375 0.7842809
## 10 2 15 0.8404059 0.7333333 0.7929208
## 11 3 15 0.8709914 0.7375000 0.7899944
## 12 5 15 0.9007186 0.7083333 0.8124303
## 13 1 10 0.7766267 0.7609375 0.7842809
## 14 2 10 0.8346440 0.7098958 0.7689521
## 15 3 10 0.8657152 0.7635417 0.8131271
## 16 5 10 0.9039188 0.6161458 0.7660256
## 17 1 5 0.7766267 0.7609375 0.7842809
## 18 2 5 0.8414039 0.7442708 0.7867893
## 19 3 5 0.8710512 0.7041667 0.7806577
## 20 5 5 0.9132535 0.7125000 0.8113155
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 10 0.8657152 0.7635417 0.8131271
##
## Observation 1 has a predicted value 0.321
## since this is the weighted average response across the 12 nodes it is a member of:
##
## 1) Node 48, containing 68 training observations, with node mean 0.324 and weight 0.274 :
## 380 <= texture_sumvariance_nondir_post1
## texture_inversediffmoment_nondir_post3 <= 0.15
## irregularity <= 0.98
##
## 2) Node 26, containing 45 training observations, with node mean 0.222 and weight 0.128 :
## irregularity <= 0.93
## 0.17 <= skew_F_r_i
## dce3SE4 <= 1.6
##
## 3) Node 34, containing 53 training observations, with node mean 0.642 and weight 0.11 :
## 480 <= texture_sumvariance_nondir_post1
## 7.1 <= V18
## 6.9 <= Tpeak_inside
##
## 4) Node 74, containing 200 training observations, with node mean 0.18 and weight 0.101 :
## irregularity <= 0.98
## texture_inversediffmoment_nondir_post3 <= 0.16
## texture_sumvariance_nondir_post1 <= 780
##
## 5) Node 71, containing 162 training observations, with node mean 0.167 and weight 0.0856 :
## irregularity <= 0.97
## dce3SE4 <= 1
##
## 6) Node 63, containing 121 training observations, with node mean 0.264 and weight 0.0784 :
## 0.92 <= irregularity <= 0.98
## 8.3 <= Tpeak_inside
##
## 7) Node 78, containing 271 training observations, with node mean 0.232 and weight 0.0676 :
## 7.8 <= Tpeak_inside
## irregularity <= 0.98
##
## 8) Node 47, containing 68 training observations, with node mean 0.456 and weight 0.0502 :
## 380 <= texture_sumvariance_nondir_post1
## 0.79 <= dce3SE12
## irregularity <= 0.96
##
## 9) Node 52, containing 74 training observations, with node mean 0.473 and weight 0.0415 :
## 5.9 <= Tpeak_inside
## irregularity <= 0.98
## 380 <= texture_sumvariance_nondir_post1
##
## 10) Node 77, containing 259 training observations, with node mean 0.292 and weight 0.0314 :
## 0.84 <= dce3SE17
## 5.7 <= Tpeak_inside
## irregularity <= 0.98
##
## 11) Node 4, containing 15 training observations, with node mean 0.4 and weight 0.022 :
## alpha_countor <= 0.38
## irregularity <= 0.98
## 1.9 <= lateSE18
##
## 12) Node 80, containing 559 training observations, with node mean 0.403 and weight 0.01 :
## ROOT NODE
##
## ============ bagging trees treedata_all
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7658749 0.7484375 0.7882102
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.8380572 0.7588542 0.8421875
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8795077 0.7192708 0.8397727
## maxinter: 5 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 4 5 25 0.9031603 0.7552083 0.8681818
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 1 20 0.7659548 0.7510417 0.7903409
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 2 20 0.8360479 0.6989583 0.8255682
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 7 3 20 0.8756487 0.7494792 0.8580966
## maxinter: 5 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 8 5 20 0.9118097 0.734375 0.85625
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 1 15 0.7647904 0.7416667 0.7894886
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 10 2 15 0.8291949 0.7322917 0.8144886
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 11 3 15 0.867811 0.7729167 0.8565341
## maxinter: 5 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 12 5 15 0.915356 0.5885417 0.7823864
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 10 0.7658749 0.7484375 0.7882102
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 10 0.8332468 0.7244792 0.8042614
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 10 0.8645243 0.7541667 0.8295455
## maxinter: 5 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 16 5 10 0.9211111 0.715625 0.8588068
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 17 1 5 0.7661344 0.75 0.7934659
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 18 2 5 0.8353227 0.7135417 0.806392
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 19 3 5 0.8726347 0.765625 0.862358
## maxinter: 5 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 20 5 5 0.9110047 0.7052083 0.8488636
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7658749 0.7484375 0.7882102
## 2 2 25 0.8380572 0.7588542 0.8421875
## 3 3 25 0.8795077 0.7192708 0.8397727
## 4 5 25 0.9031603 0.7552083 0.8681818
## 5 1 20 0.7659548 0.7510417 0.7903409
## 6 2 20 0.8360479 0.6989583 0.8255682
## 7 3 20 0.8756487 0.7494792 0.8580966
## 8 5 20 0.9118097 0.7343750 0.8562500
## 9 1 15 0.7647904 0.7416667 0.7894886
## 10 2 15 0.8291949 0.7322917 0.8144886
## 11 3 15 0.8678110 0.7729167 0.8565341
## 12 5 15 0.9153560 0.5885417 0.7823864
## 13 1 10 0.7658749 0.7484375 0.7882102
## 14 2 10 0.8332468 0.7244792 0.8042614
## 15 3 10 0.8645243 0.7541667 0.8295455
## 16 5 10 0.9211111 0.7156250 0.8588068
## 17 1 5 0.7661344 0.7500000 0.7934659
## 18 2 5 0.8353227 0.7135417 0.8063920
## 19 3 5 0.8726347 0.7656250 0.8623580
## 20 5 5 0.9110047 0.7052083 0.8488636
## maxinter nodesize rocTrain rocValid rocTest
## 4 5 25 0.9031603 0.7552083 0.8681818
##
## Observation 1 has a predicted value 0.37
## since this is the weighted average response across the 11 nodes it is a member of:
##
## 1) Node 76, containing 39 training observations, with node mean 0.718 and weight 0.201 :
## 0.92 <= irregularity <= 0.98
## UptakeRate_inside <= 0.73
## A_inside <= 1.6
## 0.99 <= lateSE8
##
## 2) Node 109, containing 138 training observations, with node mean 0.151 and weight 0.182 :
## irregularity <= 0.98
## dce2SE19 <= 0.87
## lateSE1 <= 1.2
##
## 3) Node 37, containing 17 training observations, with node mean 0.353 and weight 0.18 :
## 210 <= texture_diffvariance_nondir_post2
## irregularity <= 0.94
##
## 4) Node 94, containing 67 training observations, with node mean 0.239 and weight 0.173 :
## earlySE1 <= 1.3
## min_F_r_i <= 390
## V14 <= 23
## 1.4 <= texture_diffentropy_nondir_post4
##
## 5) Node 33, containing 16 training observations, with node mean 0.5 and weight 0.093 :
## irregularity <= 0.98
## max_F_r_i <= 1600
## -0.05 <= Kpeak_inside
## texture_inversediffmoment_nondir_post4 <= 0.19
## 200 <= texture_diffvariance_nondir_post2
##
## 6) Node 111, containing 167 training observations, with node mean 0.18 and weight 0.0651 :
## irregularity <= 0.98
## texture_inversediffmoment_nondir_post4 <= 0.13
##
## 7) Node 68, containing 31 training observations, with node mean 0.194 and weight 0.0297 :
## min_F_r_i <= 330
## 56 <= ave_T210
## irregularity <= 0.98
## texture_inversediffmoment_nondir_post4 <= 0.14
##
## 8) Node 85, containing 51 training observations, with node mean 0.275 and weight 0.0296 :
## min_F_r_i <= 330
## 56 <= ave_T210
## irregularity <= 0.98
## 95 <= ave_T21
## 92 <= T2mean_F_r_i
##
## 9) Node 107, containing 130 training observations, with node mean 0.423 and weight 0.0232 :
## UptakeRate_inside <= 0.54
## 0.93 <= irregularity <= 0.98
## 0.88 <= lateSE12
##
## 10) Node 91, containing 60 training observations, with node mean 0.567 and weight 0.0123 :
## irregularity <= 0.98
## 150 <= texture_diffvariance_nondir_post2 <= 320
##
## 11) Node 116, containing 559 training observations, with node mean 0.403 and weight 0.01 :
## ROOT NODE
## id C NC pred obs
## 1 4 0.4044551 0.5955449 NC NC
## 2 5 0.5001394 0.4998606 C C
## 3 8 0.5014587 0.4985413 C C
## 4 9 0.3956300 0.6043700 NC C
## 5 26 0.2600355 0.7399645 NC NC
## 6 27 0.5001394 0.4998606 C NC
## id C NC pred obs
## 1 4 0.5327987 0.4672013 C NC
## 2 5 0.5879184 0.4120816 C C
## 3 8 0.3151870 0.6848130 NC C
## 4 9 0.3574020 0.6425980 NC C
## 5 26 0.3980066 0.6019934 NC NC
## 6 27 0.4196659 0.5803341 NC NC
## id C NC pred obs
## 1 4 0.3210794 0.6789206 NC NC
## 2 5 0.3327871 0.6672129 NC C
## 3 8 0.3572664 0.6427336 NC C
## 4 9 0.4164816 0.5835184 NC C
## 5 26 0.2753277 0.7246723 NC NC
## 6 27 0.3378711 0.6621289 NC NC
## id C NC pred obs
## 1 4 0.3699607 0.6300393 NC NC
## 2 5 0.4716106 0.5283894 NC C
## 3 8 0.4559276 0.5440724 NC C
## 4 9 0.5954447 0.4045553 C C
## 5 26 0.2899219 0.7100781 NC NC
## 6 27 0.3533576 0.6466424 NC NC
##
## Call:
## roc.default(response = perf_imgT2$obs, predictor = perf_imgT2$C)
##
## Data: perf_imgT2$C in 416 controls (perf_imgT2$obs C) > 659 cases (perf_imgT2$obs NC).
## Area under the curve: 0.727
##
## Call:
## roc.default(response = perf_allT2$obs, predictor = perf_allT2$C)
##
## Data: perf_allT2$C in 412 controls (perf_allT2$obs C) > 663 cases (perf_allT2$obs NC).
## Area under the curve: 0.7321
##
## Call:
## roc.default(response = perf_imgT1$obs, predictor = perf_imgT1$C)
##
## Data: perf_imgT1$C in 438 controls (perf_imgT1$obs C) > 637 cases (perf_imgT1$obs NC).
## Area under the curve: 0.8312
##
## Call:
## roc.default(response = perf_all$obs, predictor = perf_all$C)
##
## Data: perf_all$C in 431 controls (perf_all$obs C) > 644 cases (perf_all$obs NC).
## Area under the curve: 0.8505
## Area under the curve: 0.727
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.3910212 0.6273 0.6707 0.7139 0.6434 0.6813 0.7178
## Area under the curve: 0.7321
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.363285 0.7791 0.8155 0.8519 0.4947 0.5324 0.5686
## Area under the curve: 0.8312
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.4158285 0.6644 0.7055 0.7489 0.7881 0.8179 0.8462
## Area under the curve: 0.8505
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.3939238 0.6914 0.7309 0.7703 0.809 0.8385 0.8665
## massB massM nonmassB nonmassM
## 218 151 131 64
## massB massM nonmassB nonmassM
## 22 17 11 13
## massB massM nonmassB nonmassM
## 218 151 131 64
## massB massM nonmassB nonmassM
## 22 17 11 13
## massB massM nonmassB nonmassM
## 218 151 131 64
## massB massM nonmassB nonmassM
## 22 17 11 13
## 0.06511628 0.05
## 0.01492537 0.05
## Selected features for group: MeanDecreaseGini imgT2
## =========NULL
## [1] "T2RGH_mean" "ave_T210"
## [3] "T2texture_correlation_nondir" "T2texture_entropy_nondir"
## [5] "T2texture_energy_nondir" "T2var_F_r_i"
## [7] "T2grad_margin_var" "ave_T213"
## [9] "T2kurt_F_r_i" "ave_T27"
## [11] "ave_T212" "T2texture_inversediffmoment_nondir"
## [13] "T2skew_F_r_i" "T2texture_sumaverage_nondir"
## [15] "T2texture_diffentropy_nondir" "ave_T25"
## [17] "ave_T21" "T2texture_diffvariance_nondir"
## [19] "ave_T211" "ave_T26"
## [21] "ave_T22" "T2texture_variance_nondir"
## [23] "ave_T218" "ave_T219"
## [25] "ave_T29" "T2RGH_var"
## [27] "ave_T216" "ave_T20"
## [29] "ave_T217" "ave_T28"
## [31] "T2mean_F_r_i" "ave_T23"
## [33] "ave_T214" "T2max_F_r_i"
## [35] "ave_T24" "ave_T215"
## [37] "T2min_F_r_i" "T2texture_sumentropy_nondir"
## [39] "T2grad_margin" "T2_lesionSI"
## 0.03414634 0.05
## Selected features for group: MeanDecreaseGini allT2
## =========NULL
## [1] "ave_T210" "T2RGH_var"
## [3] "T2texture_correlation_nondir" "T2texture_energy_nondir"
## [5] "ave_T27" "T2texture_sumaverage_nondir"
## [7] "T2kurt_F_r_i" "ave_T213"
## [9] "T2texture_diffvariance_nondir" "ave_T25"
## [11] "ave_T218" "T2texture_inversediffmoment_nondir"
## [13] "ave_T219" "ave_T216"
## [15] "ave_T211" "ave_T24"
## [17] "ave_T23" "T2_lesionSIstd"
## [19] "ave_T26" "ave_T21"
## [21] "T2texture_entropy_nondir" "ave_T214"
## [23] "T2grad_margin" "T2skew_F_r_i"
## [25] "ave_T217" "ave_T22"
## [27] "T2grad_margin_var" "LMSIR_predicted"
## [29] "T2texture_sumvariance_nondir" "ave_T215"
## [31] "ave_T29" "T2texture_sumentropy_nondir"
## [33] "T2texture_diffentropy_nondir" "ave_T28"
## [35] "T2RGH_mean" "T2min_F_r_i"
## [37] "T2wSI_predicted" "T2_lesionSI"
## 0.05031447 0.05
## 0.05298013 0.05
## -0.1398601 0.05
## Selected features for group: MeanDecreaseGini imgT1
## =========NULL
## [1] "texture_variance_nondir_post2" "SER_inside"
## [3] "mean_F_r_i" "max_RGH_var"
## [5] "V14" "Vr_post_1_countor"
## [7] "min_F_r_i" "texture_energy_nondir_post4"
## [9] "texture_inversediffmoment_nondir_post2" "texture_diffvariance_nondir_post3"
## [11] "V5" "Tpeak_countor"
## [13] "earlySE4" "iAUC1_inside"
## [15] "skew_F_r_i" "texture_sumentropy_nondir_post4"
## [17] "V7" "dce3SE17"
## [19] "washoutRate_inside" "lateSE12"
## [21] "dce3SE18" "ivVariance"
## [23] "dce3SE5" "earlySE9"
## [25] "lateSE2" "dce3SE13"
## [27] "dce3SE16" "texture_sumentropy_nondir_post3"
## [29] "lateSE19" "earlySE13"
## [31] "earlySE3" "Kpeak_inside"
## [33] "texture_diffentropy_nondir_post4" "earlySE2"
## [35] "lateSE8" "A_inside"
## [37] "V17" "texture_sumaverage_nondir_post4"
## [39] "V9" "alpha_inside"
## [41] "beta_inside"
## 0.04968944 0.05
## Selected features for group: MeanDecreaseGini all
## =========NULL
## [1] "irregularity" "texture_variance_nondir_post2"
## [3] "texture_sumaverage_nondir_post2" "earlySE8"
## [5] "texture_diffvariance_nondir_post2" "alpha_countor"
## [7] "V5" "Vr_post_1_countor"
## [9] "texture_energy_nondir_post3" "texture_diffentropy_nondir_post4"
## [11] "lateSE11" "max_RGH_var"
## [13] "ave_T210" "V6"
## [15] "V4" "dce2SE14"
## [17] "ave_T26" "Vr_increasingRate_countor"
## [19] "kurt_F_r_i" "earlySE18"
## [21] "T2texture_sumvariance_nondir" "iAUC1_countor"
## [23] "ave_T23" "lateSE13"
## [25] "dce3SE10" "lateSE2"
## [27] "ave_T218" "earlySE2"
## [29] "dce2SE7" "ave_T211"
## [31] "dce2SE0" "ave_T21"
## [33] "V17" "texture_entropy_nondir_post2"
## [35] "A_inside"
## lesion_id cad_pt_no_txt exam_a_number_txt BIRADS lesion_label
## 2 2 0016 6920252 4 massB
## 19 19 0129 5326737 4 massB
## 25 25 0133 7072006 4 massB
## 29 29 0168 5240535 4 massB
## 37 37 0186 6869828 4 nonmassM
## 53 53 0212 4734525 4 massB
## 56 56 0229 6831376 5 nonmassB
## 59 59 0246 7485590 4 massB
## 140 140 0684 5266209 4 nonmassM
## 141 141 0685 5456684 4 massB
## 151 151 0692 5199366 4 massB
## 160 160 0713 5150291 5 massM
## 161 161 0713 5150291 5 nonmassM
## 197 197 0757 4779344 4 nonmassB
## 201 201 0765 5094113 4 nonmassB
## 220 220 0791 5365218 5 massM
## 221 221 0791 5365218 5 nonmassM
## 231 231 0805 5059167 4 nonmassB
## 238 238 0814 4704240 5 massM
## 239 239 0814 6667547 4 nonmassB
## 259 259 0846 4800867 5 massM
## 264 264 0851 4593282 4 massB
## 265 265 0851 4593282 4 massM
## 310 310 0920 7095635 4 massB
## 312 312 0924 7532614 4 massB
## 313 313 0924 7532614 4 massB
## 314 314 0934 5314924 4 nonmassB
## 330 330 0978 4851428 4 nonmassB
## 336 336 0999 6925971 3 massB
## 337 337 1003 6682777 4 nonmassM
## 344 344 1018 4773924 4 nonmassB
## 392 392 2049 5458850 5 nonmassM
## 393 393 2049 5458850 5 massM
## 405 405 2073 4745825 5 massM
## 406 406 2073 4745825 5 nonmassM
## 410 410 2079 4591198 3 massB
## 440 440 3053 7041483 6 massB
## 441 441 3053 7449310 5 nonmassM
## 450 450 3063 7053508 6 nonmassM
## 464 464 3080 7033654 6 nonmassM
## 467 467 3082 5355166 6 massB
## 468 468 3082 5355166 6 massB
## 469 469 3082 7080675 4 nonmassB
## 512 512 6008 4644038 6 massM
## 520 520 6020 ACC109177 6 massM
## 522 522 6022 5046558 4 nonmassB
## 523 523 6022 5046558 6 massM
## 526 526 6024 5008021 5 massM
## 527 527 6024 5008021 5 nonmassM
## 528 528 6025 5111910 6 massM
## 529 529 6025 5111910 6 massM
## 530 530 6025 5111910 6 nonmassM
## 568 568 6050 5225817 4 nonmassB
## 594 594 7011 6918051 4 massB
## 595 595 7018 6803089 4 massM
## 596 596 7018 7138226 2 massM
## 599 599 7030 7538617 4 massB
## 600 600 7030 7538617 4 massB
## 608 608 7077 5077480 5 massM
## 609 609 7077 5077480 5 massM
## 610 610 7077 5077480 5 nonmassM
## 629 629 7189 7068978 4 massB
## 630 630 7189 7068978 4 massB
## lesion_diagnosis find_t2_signal_int
## 2 FLORID DUCT HYPERPLASIA Slightly hyperintense
## 19 BENIGN BREAST TISSUE Hyperintense
## 25 FIBROCYSTIC Hypointense or not seen
## 29 FIBROCYSTIC None
## 37 InsituDuctal None
## 53 FIBROADENOMA Hyperintense
## 56 FIBROCYSTIC None
## 59 BENIGN BREAST TISSUE Hyperintense
## 140 InsituDuctal None
## 141 FIBROCYSTIC None
## 151 FIBROADENOMA Hypointense or not seen
## 160 InvasiveDuctal None
## 161 InvasiveDuctal None
## 197 ATYPICAL DUCTAL HYPERPLASIA Hyperintense
## 201 ATYPICAL DUCTAL HYPERPLASIA None
## 220 InvasiveLobular None
## 221 InvasiveLobular None
## 231 FIBROCYSTIC None
## 238 InvasiveDuctal None
## 239 ATYPICAL DUCTAL HYPERPLASIA None
## 259 MetaplasticCarcinoma Hypointense or not seen
## 264 InsituLobular Hyperintense
## 265 InvasiveDuctal Hyperintense
## 310 FIBROADENOMA Hypointense or not seen
## 312 ADENOSIS Hyperintense
## 313 ADENOSIS Hypointense or not seen
## 314 FIBROADENOMA None
## 330 FIBROCYSTIC None
## 336 FIBROADENOMA Slightly hyperintense
## 337 InvasiveDuctal Slightly hyperintense
## 344 BENIGN BREAST TISSUE Hypointense or not seen
## 392 InvasiveDuctal Hypointense or not seen
## 393 InvasiveDuctal Hypointense or not seen
## 405 InvasiveDuctal micropapillary None
## 406 InvasiveLobular None
## 410 benign lymphoid tissue None
## 440 COLUMNAR CELL CHANGES Hypointense or not seen
## 441 InvasiveDuctal None
## 450 LYMPHOVASCULAR INVASION None
## 464 InsituDuctal Hypointense or not seen
## 467 FIBROADENOMA None
## 468 FIBROCYSTIC None
## 469 FIBROCYSTIC None
## 512 InvasiveDuctal Hypointense or not seen
## 520 InvasiveDuctal Hypointense or not seen
## 522 FIBROADENOMA None
## 523 InvasiveDuctal None
## 526 InvasiveDuctal None
## 527 InvasiveDuctal None
## 528 InvasiveDuctal Hyperintense
## 529 InvasiveDuctal Slightly hyperintense
## 530 InvasiveDuctal None
## 568 FIBROSIS Hypointense or not seen
## 594 BENIGN BREAST TISSUE Hypointense or not seen
## 595 InsituDuctal None
## 596 InsituDuctal None
## 599 BENIGN BREAST TISSUE Slightly hyperintense
## 600 BENIGN BREAST TISSUE Hypointense or not seen
## 608 InsituDuctal None
## 609 InsituDuctal None
## 610 InsituDuctal None
## 629 FIBROADENOMA Hyperintense
## 630 BENIGN BREAST TISSUE Hyperintense
##
## ============ bagging trees treedata_imgT2
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6708136 0.6828283 0.621802
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.7401279 0.6636364 0.6383482
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.79291 0.6464646 0.692436
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 4 1 20 0.6710135 0.6828283 0.6220801
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 2 20 0.7405011 0.6868687 0.6511402
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 3 20 0.7922703 0.6712121 0.6758899
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 7 1 15 0.6708136 0.6828283 0.621802
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 8 2 15 0.7528753 0.6818182 0.6408509
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 3 15 0.8102752 0.6515152 0.6833982
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 10 1 10 0.6710135 0.6828283 0.6220801
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 11 2 10 0.7537149 0.6909091 0.6662959
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 12 3 10 0.7828747 0.6454545 0.676168
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 5 0.6710135 0.6828283 0.6220801
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 5 0.7572266 0.7065657 0.6567019
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 5 0.7864663 0.7282828 0.7246941
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6708136 0.6828283 0.6218020
## 2 2 25 0.7401279 0.6636364 0.6383482
## 3 3 25 0.7929100 0.6464646 0.6924360
## 4 1 20 0.6710135 0.6828283 0.6220801
## 5 2 20 0.7405011 0.6868687 0.6511402
## 6 3 20 0.7922703 0.6712121 0.6758899
## 7 1 15 0.6708136 0.6828283 0.6218020
## 8 2 15 0.7528753 0.6818182 0.6408509
## 9 3 15 0.8102752 0.6515152 0.6833982
## 10 1 10 0.6710135 0.6828283 0.6220801
## 11 2 10 0.7537149 0.6909091 0.6662959
## 12 3 10 0.7828747 0.6454545 0.6761680
## 13 1 5 0.6710135 0.6828283 0.6220801
## 14 2 5 0.7572266 0.7065657 0.6567019
## 15 3 5 0.7864663 0.7282828 0.7246941
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 5 0.7864663 0.7282828 0.7246941
##
## Observation 1 has a predicted value 0.293
## since this is the weighted average response across the 7 nodes it is a member of:
##
## 1) Node 52, containing 344 training observations, with node mean 0.308 and weight 0.309 :
## T2texture_entropy_nondir <= 3.5
## 39 <= ave_T210
## 76 <= ave_T213
##
## 2) Node 49, containing 252 training observations, with node mean 0.424 and weight 0.225 :
## 0.00053 <= T2texture_energy_nondir
## 280 <= T2RGH_var
## 56 <= ave_T210
##
## 3) Node 37, containing 81 training observations, with node mean 0.155 and weight 0.158 :
## 9.5 <= T2min_F_r_i
## ave_T213 <= 220
##
## 4) Node 43, containing 139 training observations, with node mean 0.288 and weight 0.146 :
## T2texture_diffentropy_nondir <= 1.5
## T2texture_energy_nondir <= 0.0038
##
## 5) Node 40, containing 109 training observations, with node mean 0.193 and weight 0.12 :
## T2RGH_mean <= 22
##
## 6) Node 48, containing 251 training observations, with node mean 0.275 and weight 0.0323 :
## 0.00048 <= T2texture_energy_nondir
## T2texture_correlation_nondir <= 0.26
## 54 <= T2mean_F_r_i
##
## 7) Node 53, containing 560 training observations, with node mean 0.381 and weight 0.01 :
## ROOT NODE
##
## ============ bagging trees treedata_allT2
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6879856 0.7151515 0.700341
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.7511095 0.7575758 0.7600171
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8130939 0.7030303 0.7468031
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 4 1 20 0.6879856 0.7151515 0.700341
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 2 20 0.753555 0.7030303 0.7321682
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 3 20 0.8163324 0.6813131 0.7627167
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 7 1 15 0.6879856 0.7151515 0.700341
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 8 2 15 0.7456587 0.7010101 0.7429668
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 3 15 0.8067835 0.6747475 0.7296107
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 10 1 10 0.6879856 0.7151515 0.700341
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 11 2 10 0.7417405 0.7272727 0.7389883
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 12 3 10 0.8017192 0.7045455 0.7617221
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 5 0.6879856 0.7151515 0.700341
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 5 0.7519291 0.7212121 0.7485081
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 5 0.8075232 0.6888889 0.7685422
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.6879856 0.7151515 0.7003410
## 2 2 25 0.7511095 0.7575758 0.7600171
## 3 3 25 0.8130939 0.7030303 0.7468031
## 4 1 20 0.6879856 0.7151515 0.7003410
## 5 2 20 0.7535550 0.7030303 0.7321682
## 6 3 20 0.8163324 0.6813131 0.7627167
## 7 1 15 0.6879856 0.7151515 0.7003410
## 8 2 15 0.7456587 0.7010101 0.7429668
## 9 3 15 0.8067835 0.6747475 0.7296107
## 10 1 10 0.6879856 0.7151515 0.7003410
## 11 2 10 0.7417405 0.7272727 0.7389883
## 12 3 10 0.8017192 0.7045455 0.7617221
## 13 1 5 0.6879856 0.7151515 0.7003410
## 14 2 5 0.7519291 0.7212121 0.7485081
## 15 3 5 0.8075232 0.6888889 0.7685422
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 5 0.8075232 0.6888889 0.7685422
##
## Observation 1 has a predicted value 0.284
## since this is the weighted average response across the 8 nodes it is a member of:
##
## 1) Node 23, containing 0.5 training observations, with node mean 0.0857 and weight 0.279 :
## T2wSI_predicted in {Hyperintense,Hypointense or not seen,Slightly hyperintense}
## T2RGH_mean <= 22
##
## 2) Node 39, containing 203 training observations, with node mean 0.421 and weight 0.21 :
## 350 <= T2RGH_var
## 130 <= ave_T211
##
## 3) Node 44, containing 362 training observations, with node mean 0.38 and weight 0.163 :
## T2texture_entropy_nondir <= 3.5
## T2texture_diffentropy_nondir <= 1.8
## 56 <= ave_T210
##
## 4) Node 40, containing 0.5 training observations, with node mean 0.291 and weight 0.152 :
## T2wSI_predicted in {Hyperintense,Hypointense or not seen,Slightly hyperintense}
## 0.1 <= T2texture_correlation_nondir
## 53 <= T2_lesionSI
##
## 5) Node 41, containing 230 training observations, with node mean 0.317 and weight 0.136 :
## T2texture_entropy_nondir <= 3.5
## T2texture_diffentropy_nondir <= 1.5
##
## 6) Node 43, containing 241 training observations, with node mean 0.414 and weight 0.0347 :
## T2texture_entropy_nondir <= 3.5
## 280 <= T2RGH_var
## 56 <= ave_T210
##
## 7) Node 42, containing 0.5 training observations, with node mean 0.273 and weight 0.0154 :
## 56 <= ave_T210
## T2wSI_predicted in {Hyperintense,Hypointense or not seen,Slightly hyperintense}
## 56 <= ave_T27
##
## 8) Node 45, containing 563 training observations, with node mean 0.381 and weight 0.01 :
## ROOT NODE
##
## ============ bagging trees treedata_imgT1
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7617112 0.8383838 0.7724585
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.8151796 0.8520202 0.8009012
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8615646 0.8843434 0.8445508
## maxinter: 5 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 4 5 25 0.9135337 0.8565657 0.8667981
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 1 20 0.7617112 0.8383838 0.7724585
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 2 20 0.8201906 0.8378788 0.8066742
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 7 3 20 0.8523889 0.8449495 0.8199099
## maxinter: 5 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 8 5 20 0.9113414 0.8378788 0.8753872
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 1 15 0.7617112 0.8383838 0.7724585
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 10 2 15 0.8161791 0.8141414 0.7861166
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 11 3 15 0.8580263 0.870202 0.8325824
## maxinter: 5 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 12 5 15 0.9132405 0.8040404 0.8400451
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 10 0.7617112 0.8383838 0.7724585
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 10 0.8194576 0.8414141 0.7987891
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 10 0.8606717 0.879798 0.8569417
## maxinter: 5 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 16 5 10 0.8986273 0.8449495 0.881301
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 17 1 5 0.7617112 0.8383838 0.7724585
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 18 2 5 0.8199773 0.8474747 0.7930161
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 19 3 5 0.8609516 0.8383838 0.8189242
## maxinter: 5 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 20 5 5 0.9085893 0.8974747 0.8869333
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7617112 0.8383838 0.7724585
## 2 2 25 0.8151796 0.8520202 0.8009012
## 3 3 25 0.8615646 0.8843434 0.8445508
## 4 5 25 0.9135337 0.8565657 0.8667981
## 5 1 20 0.7617112 0.8383838 0.7724585
## 6 2 20 0.8201906 0.8378788 0.8066742
## 7 3 20 0.8523889 0.8449495 0.8199099
## 8 5 20 0.9113414 0.8378788 0.8753872
## 9 1 15 0.7617112 0.8383838 0.7724585
## 10 2 15 0.8161791 0.8141414 0.7861166
## 11 3 15 0.8580263 0.8702020 0.8325824
## 12 5 15 0.9132405 0.8040404 0.8400451
## 13 1 10 0.7617112 0.8383838 0.7724585
## 14 2 10 0.8194576 0.8414141 0.7987891
## 15 3 10 0.8606717 0.8797980 0.8569417
## 16 5 10 0.8986273 0.8449495 0.8813010
## 17 1 5 0.7617112 0.8383838 0.7724585
## 18 2 5 0.8199773 0.8474747 0.7930161
## 19 3 5 0.8609516 0.8383838 0.8189242
## 20 5 5 0.9085893 0.8974747 0.8869333
## maxinter nodesize rocTrain rocValid rocTest
## 20 5 5 0.9085893 0.8974747 0.8869333
##
## Observation 1 has a predicted value 0.205
## since this is the weighted average response across the 12 nodes it is a member of:
##
## 1) Node 136, containing 276 training observations, with node mean 0.199 and weight 0.228 :
## texture_variance_nondir_post2 <= 180
## alpha_inside <= 0.64
## 370 <= mean_F_r_i
##
## 2) Node 134, containing 235 training observations, with node mean 0.191 and weight 0.183 :
## alpha_inside <= 0.62
## lateSE8 <= 1.8
## mean_F_r_i <= 930
## Vr_post_1_countor <= 0.31
##
## 3) Node 125, containing 98 training observations, with node mean 0.235 and weight 0.141 :
## washoutRate_inside <= 0.0087
## 0.69 <= SER_inside
## Kpeak_inside <= -0.00027
## mean_F_r_i <= 1100
## texture_diffvariance_nondir_post3 <= 180
##
## 4) Node 45, containing 20 training observations, with node mean 0.15 and weight 0.0986 :
## SER_inside <= 0.79
## Vr_post_1_countor <= 0.13
## V17 <= 8.1
## 560 <= iAUC1_inside
##
## 5) Node 133, containing 197 training observations, with node mean 0.155 and weight 0.0902 :
## earlySE9 <= 1.6
## SER_inside <= 0.77
## 150 <= min_F_r_i
## -0.054 <= Kpeak_inside
## skew_F_r_i <= 0.68
##
## 6) Node 135, containing 270 training observations, with node mean 0.196 and weight 0.0841 :
## SER_inside <= 0.8
## texture_variance_nondir_post2 <= 300
## beta_inside <= 0.12
##
## 7) Node 121, containing 86 training observations, with node mean 0.337 and weight 0.0759 :
## alpha_inside <= 0.56
## texture_sumentropy_nondir_post4 <= 2
## V14 <= 21
## texture_diffentropy_nondir_post4 <= 1.5
## 0.036 <= beta_inside
##
## 8) Node 132, containing 176 training observations, with node mean 0.239 and weight 0.0489 :
## V7 <= 7.4
##
## 9) Node 131, containing 170 training observations, with node mean 0.165 and weight 0.0179 :
## mean_F_r_i <= 760
## washoutRate_inside <= 0.0044
## 1e-04 <= beta_inside
## 140 <= min_F_r_i
##
## 10) Node 127, containing 116 training observations, with node mean 0.0763 and weight 0.0144 :
## texture_variance_nondir_post2 <= 180
## SER_inside <= 0.72
## 0.45 <= earlySE9
## skew_F_r_i <= 0.78
##
## 11) Node 137, containing 564 training observations, with node mean 0.381 and weight 0.01 :
## ROOT NODE
##
## 12) Node 130, containing 168 training observations, with node mean 0.185 and weight 0.00728 :
## texture_variance_nondir_post2 <= 310
## texture_inversediffmoment_nondir_post2 <= 0.16
## V17 <= 30
##
## ============ bagging trees treedata_all
## maxinter: 1 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7722929 0.8671717 0.835578
## maxinter: 2 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 2 2 25 0.8292997 0.8686869 0.8734568
## maxinter: 3 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 3 3 25 0.8640634 0.8777778 0.9013749
## maxinter: 5 nodesize: 25
## maxinter nodesize rocTrain rocValid rocTest
## 4 5 25 0.9180649 0.8424242 0.884119
## maxinter: 1 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 5 1 20 0.7722929 0.8671717 0.835578
## maxinter: 2 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 6 2 20 0.8337709 0.9060606 0.9006734
## maxinter: 3 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 7 3 20 0.8625175 0.8565657 0.8757015
## maxinter: 5 nodesize: 20
## maxinter nodesize rocTrain rocValid rocTest
## 8 5 20 0.910062 0.8909091 0.9147026
## maxinter: 1 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 9 1 15 0.7722929 0.8671717 0.835578
## maxinter: 2 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 10 2 15 0.8303392 0.8929293 0.8901515
## maxinter: 3 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 11 3 15 0.8542613 0.8969697 0.9008137
## maxinter: 5 nodesize: 15
## maxinter nodesize rocTrain rocValid rocTest
## 12 5 15 0.8933231 0.8464646 0.8880471
## maxinter: 1 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 13 1 10 0.7722929 0.8671717 0.835578
## maxinter: 2 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 14 2 10 0.8251016 0.8893939 0.8702301
## maxinter: 3 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 15 3 10 0.8626241 0.879798 0.8835578
## maxinter: 5 nodesize: 10
## maxinter nodesize rocTrain rocValid rocTest
## 16 5 10 0.8970947 0.8651515 0.8914141
## maxinter: 1 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 17 1 5 0.7722929 0.8671717 0.835578
## maxinter: 2 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 18 2 5 0.8303925 0.8888889 0.8755612
## maxinter: 3 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 19 3 5 0.856407 0.8722222 0.8782267
## maxinter: 5 nodesize: 5
## maxinter nodesize rocTrain rocValid rocTest
## 20 5 5 0.9026254 0.8752525 0.9096521
## maxinter nodesize rocTrain rocValid rocTest
## 1 1 25 0.7722929 0.8671717 0.8355780
## 2 2 25 0.8292997 0.8686869 0.8734568
## 3 3 25 0.8640634 0.8777778 0.9013749
## 4 5 25 0.9180649 0.8424242 0.8841190
## 5 1 20 0.7722929 0.8671717 0.8355780
## 6 2 20 0.8337709 0.9060606 0.9006734
## 7 3 20 0.8625175 0.8565657 0.8757015
## 8 5 20 0.9100620 0.8909091 0.9147026
## 9 1 15 0.7722929 0.8671717 0.8355780
## 10 2 15 0.8303392 0.8929293 0.8901515
## 11 3 15 0.8542613 0.8969697 0.9008137
## 12 5 15 0.8933231 0.8464646 0.8880471
## 13 1 10 0.7722929 0.8671717 0.8355780
## 14 2 10 0.8251016 0.8893939 0.8702301
## 15 3 10 0.8626241 0.8797980 0.8835578
## 16 5 10 0.8970947 0.8651515 0.8914141
## 17 1 5 0.7722929 0.8671717 0.8355780
## 18 2 5 0.8303925 0.8888889 0.8755612
## 19 3 5 0.8564070 0.8722222 0.8782267
## 20 5 5 0.9026254 0.8752525 0.9096521
## maxinter nodesize rocTrain rocValid rocTest
## 8 5 20 0.910062 0.8909091 0.9147026
##
## Observation 1 has a predicted value 0.157
## since this is the weighted average response across the 9 nodes it is a member of:
##
## 1) Node 105, containing 137 training observations, with node mean 0.131 and weight 0.207 :
## irregularity <= 0.93
## 120 <= ave_T26
##
## 2) Node 102, containing 116 training observations, with node mean 0.172 and weight 0.17 :
## Vr_increasingRate_countor <= 0.098
## irregularity <= 0.98
## 1.6 <= iAUC1_countor
## texture_sumaverage_nondir_post2 <= 260
## alpha_countor <= 0.47
##
## 3) Node 82, containing 63 training observations, with node mean 0.206 and weight 0.138 :
## earlySE8 <= 0.96
## Vr_increasingRate_countor <= 0.099
## 0.4 <= dce2SE14
## 0.084 <= max_RGH_var
## Vr_post_1_countor <= 0.11
##
## 4) Node 88, containing 73 training observations, with node mean 0.0959 and weight 0.138 :
## texture_variance_nondir_post2 <= 290
## earlySE8 <= 0.96
## texture_sumaverage_nondir_post2 <= 250
## 1.2 <= dce3SE10
##
## 5) Node 109, containing 181 training observations, with node mean 0.155 and weight 0.136 :
## texture_variance_nondir_post2 <= 300
## irregularity <= 0.93
## alpha_countor <= 0.56
##
## 6) Node 106, containing 137 training observations, with node mean 0.153 and weight 0.101 :
## irregularity <= 0.97
## Vr_post_1_countor <= 0.086
## alpha_countor <= 1.2
##
## 7) Node 89, containing 72 training observations, with node mean 0.151 and weight 0.0577 :
## irregularity <= 0.93
## Vr_increasingRate_countor <= 0.56
## 63 <= T2texture_sumvariance_nondir
## V5 <= 11
## 1.1 <= lateSE11
##
## 8) Node 97, containing 101 training observations, with node mean 0.228 and weight 0.0423 :
## texture_variance_nondir_post2 <= 180
## alpha_countor <= 0.38
## texture_sumaverage_nondir_post2 <= 250
## 4.3 <= iAUC1_countor
##
## 9) Node 110, containing 564 training observations, with node mean 0.381 and weight 0.01 :
## ROOT NODE
## id C NC pred obs
## 1 2 0.2928316 0.7071684 NC NC
## 2 19 0.3054307 0.6945693 NC NC
## 3 25 0.2516258 0.7483742 NC NC
## 4 29 0.4837723 0.5162277 NC NC
## 5 37 0.3676575 0.6323425 NC C
## 6 53 0.2874368 0.7125632 NC NC
## id C NC pred obs
## 1 2 0.2840275 0.7159725 NC NC
## 2 19 0.3042624 0.6957376 NC NC
## 3 25 0.3415658 0.6584342 NC NC
## 4 29 0.4498795 0.5501205 NC NC
## 5 37 0.4709251 0.5290749 NC C
## 6 53 0.2236388 0.7763612 NC NC
## id C NC pred obs
## 1 2 0.2053191 0.7946809 NC NC
## 2 19 0.2307681 0.7692319 NC NC
## 3 25 0.2528990 0.7471010 NC NC
## 4 29 0.2026502 0.7973498 NC NC
## 5 37 0.4218565 0.5781435 NC C
## 6 53 0.1680793 0.8319207 NC NC
## id C NC pred obs
## 1 2 0.1569013 0.8430987 NC NC
## 2 19 0.2910408 0.7089592 NC NC
## 3 25 0.2932646 0.7067354 NC NC
## 4 29 0.1331342 0.8668658 NC NC
## 5 37 0.6286339 0.3713661 C C
## 6 53 0.2231512 0.7768488 NC NC
##
## Call:
## roc.default(response = perf_imgT2$obs, predictor = perf_imgT2$C)
##
## Data: perf_imgT2$C in 474 controls (perf_imgT2$obs C) > 721 cases (perf_imgT2$obs NC).
## Area under the curve: 0.7252
##
## Call:
## roc.default(response = perf_allT2$obs, predictor = perf_allT2$C)
##
## Data: perf_allT2$C in 463 controls (perf_allT2$obs C) > 732 cases (perf_allT2$obs NC).
## Area under the curve: 0.7331
##
## Call:
## roc.default(response = perf_imgT1$obs, predictor = perf_imgT1$C)
##
## Data: perf_imgT1$C in 491 controls (perf_imgT1$obs C) > 704 cases (perf_imgT1$obs NC).
## Area under the curve: 0.836
##
## Call:
## roc.default(response = perf_all$obs, predictor = perf_all$C)
##
## Data: perf_all$C in 485 controls (perf_all$obs C) > 710 cases (perf_all$obs NC).
## Area under the curve: 0.8574
## Area under the curve: 0.7252
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.3910212 0.6139 0.6582 0.7004 0.6546 0.6893 0.7226
## Area under the curve: 0.7331
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.3629964 0.7732 0.8121 0.8467 0.5109 0.5464 0.5847
## Area under the curve: 0.836
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.4156964 0.6701 0.7088 0.7475 0.8026 0.8295 0.8565
## Area under the curve: 0.8574
## 95% CI (2000 stratified bootstrap replicates):
## thresholds sp.low sp.median sp.high se.low se.median se.high
## 0.3939238 0.699 0.7381 0.7732 0.8141 0.8423 0.869
